Translational development of an SGLT2-targeted near-infrared contrast agent for intraoperative imaging of early-stage lung cancer

Abstract

Background

Non-small cell lung cancer (NSCLC) is increasingly diagnosed at early stages, yet intraoperative localization of small and subsolid lesions remains challenging. Intraoperative molecular imaging (IMI) using tumor-targeted tracers is effective in many cancers; however, no available optical agents are specific for precursor or early-stage NSCLC. Sodium-glucose cotransporter 2 (SGLT2) is upregulated in in situ and minimally invasive lung cancers. This study aimed to develop and validate GlucoGlo, a novel SGLT2-targeted near-infrared (NIR) contrast agent, for IMI in early-stage NSCLC.

Results

GlucoGlo, an SGLT2-targeted fluorescent tracer, was developed and assessed for specificity and efficacy in targeting and detecting NSCLC. Its SGLT2-specific binding was assessed by fluorescence imaging in in vitro and in vivo murine NSCLC models. Mice pretreated with an SGLT2 inhibitor were used to confirm on target binding. GlucoGlo’s performance in detecting residual tumor after resection was compared to conventional visualization and palpation. GlucoGlo selectively bound SGLT2-expressing NSCLC cell lines in vitro with minimal fluorescent signal seen in negative controls. In vivo, it accumulated in flank xenografts at clinically relevant doses (peak signal to background ratio (SBR) of 6.85 at 48 h) without toxicity. Fluorescent signal was eliminated with pretreatment with a non-fluorescent SGLT2 inhibitor, confirming target specificity. Histopathologic analysis further confirmed the selective tumor accumulation. In a murine partial-resection model, GlucoGlo exhibited significantly greater sensitivity for detecting residual tumor compared to conventional visualization and palpation in resection models (100% vs. 62.5%; p < 0.01). In ex-vivo human lung tissue, GlucoGlo accurately identified pulmonary malignancy and had significantly greater mean fluorescence in tumor areas compared to normal lung (25,218 vs. 3,371 a.u., respectively, SBR: 7.57; p = 0.009).

Conclusion

GlucoGlo demonstrated high sensitivity and specificity for SGLT2 in preclinical models of NSCLC, supporting its potential for clinical translation in intraoperative detection of ground glass opacities and early-stage lung cancer.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13550-026-01425-9.

Introduction

Lung cancer is increasingly identified at early stages due to advances in screening protocols and increased surveillance with computed tomography (CT) [1]. In 2024 alone, nearly 50,000 Americans were diagnosed with Stage I non-small cell lung cancer (NSCLC) [2], presenting a critical opportunity for curative intervention. Surgical excision remains the optimal treatment approach; however, accurately localizing and resecting small or subsolid tumors during surgery is often challenging. As a result, 10% of patients undergoing minimally invasive lung resection require conversion to an open procedure and 19% of patients with Stage I NSCLC experience a recurrence within 5 years, often due to incomplete tumor removal [3–6].

Intraoperative molecular imaging (IMI) has emerged as a powerful tool to enhance tumor visualization and improve surgical precision. IMI has shown clinical benefit in multiple malignancies, including gliomas [7], breast cancer [8], ovarian cancer [9], and lung cancer [10]. However, no FDA-approved imaging agents currently exist for detecting early-stage lung cancers, highlighting a critical unmet need in thoracic oncology.

Sodium glucose cotransporter 2 (SGLT2) is a glucose transporter that has been shown to be upregulated in early lung adenocarcinomas (LUAD). Aberrant glucose metabolism is a hallmark of neoplastic transformation, characterized by increased glycolysis even in the presence of oxygen and known as the Warburg Effect [11–13]. SGLT2 is highly expressed in early-stage LUAD, particularly in situ and minimally invasive adenocarcinomas that often appear as ground glass opacities (GGO) on CT imaging [14–16]. SGLT2 expression is minimal in normal lung tissue, making it a highly selective target for early LUAD detection and a promising target for IMI of early lung cancers [17].

In this study, we evaluated GlucoGlo, an SGLT2-targeted near-infrared (NIR) IMI agent, for the detection of early LUAD. GlucoGlo consists of a modified SGLT2 inhibitor covalently linked to indocyanine green (ICG), a clinically available NIR fluorophore detectable by standard surgical imaging devices. In this study, we conducted a comprehensive preclinical evaluation of GlucoGlo, using in vitro and animal NSCLC models and ex-vivo human lung cancers. Our findings demonstrate that GlucoGlo selectively targets SGLT2-expressing tumors with minimal background fluorescence, significantly enhancing intraoperative visualization in preclinical models. These results highlight GlucoGlo’s potential to improve surgical precision in patients with early-stage lung cancer.

Materials and methods

Study drug

GlucoGlo (C₈₀H₈₄ClN₃O₁₉S₄; 1606.34 Da) is a modified SGLT2 inhibitor conjugated to ICG (Supplemental Fig. 1). The SGLT2-inhibitor ligand, manufactured under Good Manufacturing Practices by Astra-Zeneca (Hyderabad, India), was minimally modified to facilitate conjugation, as described in Lansdell et al. [18] The excitation and emission peaks are 805 and 838 nm, respectively [19]. GlucoGlo was stored at − 20 °C and diluted with saline, phosphate-buffered saline (PBS) or culture media before use.

Fig. 1.

GlucoGlo labels lung cell lines in an SGLT2-dependent manner. a) Flow cytometry tracings of cell lines stained with anti-SGLT2 antibody (red) vs. unstained controls (blue). b) Fluorescence microscopy of cell lines incubated with GlucoGlo (red) and counterstained with DAPI (blue); 10x magnification with 20x inset

In vitro studies

Human NSCLC cell lines (H1299 and H1666) and lung fibroblasts (HD28) were obtained from American Type Culture Collection (Manassas, VA) and cultured using standard growth media (RPMI1640 media supplemented with 10% fetal bovine serum (FBS), 1% glutamine, and 1% penicillin/streptomycin) at 37 °C in 5% CO₂. Cell lines were routinely tested and maintained negative for Mycoplasma spp.

SGLT2 expression was characterized in the cell lines by flow cytometry following incubation with anti-SGLT2 monoclonal antibody (ab58298, Abcam, Cambridge, MA) and Alexa Fluor 488-conjugated secondary antibody (ab150113, Abcam, Cambridge, MA). Flow cytometry was performed on an LSR Fortessa X-20 flow cytometer (BD Biosciences, San Diego, CA) and analyzed using FlowJo software (Ashland, OR).

GlucoGlo binding and fluorescence were assessed in the same cell lines to evaluate sensitivity and specificity of the dye for SGLT2. Cells were seeded in poly-L-lysine coated chamber slides (Thermo Fisher Scientific, Somerset, NJ) with standard growth media for 24 h, then incubated with 10µM GlucoGlo for 1 h. Slides were mounted with ProLong Gold Antifade Reagent containing DAPI (Fisher Scientific, Waltham, MA) and imaged with a Leica DM6 B fluorescence microscope (Leica Microsystems, Wetzlar, Germany). Fluorescence was correlated with SGLT2 expression in the cell lines. Specificity of GlucoGlo binding was confirmed in the highest expression cell line, H1299, by competitive inhibition with unconjugated dapagliflozin. H1299 cells were seeded in the fashion previously described and treated with 10 µM GlucoGlo in the presence of 100 μm unconjugated dapagliflozin. The same staining protocol was maintained, and GlucoGlo fluorescence was imaged using the Leica microscope.

Small animal tumor model and imaging

Four-week-old female nude athymic mice (Taconic Biosciences, Germantown, NY) were housed in pathogen-free conditions. The Animal Care and Use Committee of the University of Pennsylvania approved all animal study protocols, and experiments were conducted in compliance with the Guide for the Care and Use of Laboratory Animals. Flank xenografts were established by injecting 2 × 10⁶ H1299 cells in 1:1 PBS and Matrigel (Corning Life Sciences, Corning, NY) subcutaneously.

Mice bearing palpable flank xenografts were administered GlucoGlo at varying doses (0.01, 0.05, 0.1, and 0.5 mg/kg; n = 3 per dose) via tail vein injection. Tumor fluorescence was recorded 3 to 120 h post-injection using the 800-nm channel of the Pearl Trilogy In Vivo Imaging System (LI-COR Biosciences, Lincoln, NE). Signal-to-background ratios (SBRs) were calculated by comparing mean fluorescence in tumors areas to surrounding soft tissue. Biodistribution was assessed by ex vivo organ imaging 48 h following GlucoGlo injection (0.05 mg/kg).

To evaluate safety profile of GlucoGlo, mice received a single injection of GlucoGlo (0.5 mg/kg, n = 3) or saline control (n = 3) and were monitored daily for body weight, body condition score (BCS), and local injection site reactions. Body weight over time and percent weight change were analyzed using a mixed-effects model with repeated measures.

Competitive inhibition studies were performed to evaluate SGLT2-specific binding of the dye in vivo by pretreating mice with H1299-flank tumors with 5 mg/kg free SGLT2 inhibitor in 100µL saline or 100µL saline placebo for two days before 0.05 mg/kg GlucoGlo injection (n = 4 per group). 72 h after GlucoGlo administration, mice were euthanized, imaged, and SBRs were compared between groups.

Histopathologic and fluorescent microscopic review of specimens

To understand GlucoGlo tumor accumulation at the microscopic level, xenografts resected 72 h post-injection were formalin-fixed, paraffin-embedded (FFPE), and sectioned at 5 μm and underwent hematoxylin and eosin (H&E) staining and IHC with an anti-SGLT2 monoclonal antibody (1:1000 dilution, ab58298, Abcam, Cambridge, MA, USA).

Fluorescence microscopy was performed on unstained slides of FFPE tumor specimens with an Odyssey microscopic scanner (LiCOR, Lincoln, NE) using the 800-nm channel. Areas of GlucoGlo fluorescence were correlated with tumor areas on H&E and with SGLT2 staining on IHC by a board-certified pulmonary pathologist.

Tumor margin assessment

To evaluate GlucoGlo’s utility in identifying positive and negative surgical margins, 49 mice with flank tumors received 0.05 mg/kg GlucoGlo and underwent a partial or complete tumor resection at 72 h after injection [20]. In the partial resection arm, approximately 1–2 mm of residual tumor was left in the soft tissue to simulate a positive margin. In the complete resection arm, the entire tumor was resected with wide negative margins. A blinded investigator assessed the surgical bed first using conventional visualization and palpation techniques and recorded their assessment of margin status. The surgical bed was then imaged using the Pearl Trilogy Imaging System, after which the blinded investigator revised their initial assessment if needed. Positive margins were confirmed via H&E staining.

Imaging of human lung cancers using precision-cut lung slices (PCLS)

We performed ex vivo incubation of the tracer with freshly resected patient specimens to test the translational potential of the tracer. Patients with lung cancer undergoing surgical resection provided informed consent for tissue donation. The study was performed in line with the principles of the Declaration of Helsinki and was approved by the University of Pennsylvania Institutional Review Board.

2cm³ specimens of tumor and matched distant normal parenchyma (n = 4) were obtained from each patient. The tissue was inflated with 10% Low-Melt Agarose (Fisher Scientific, PA, USA), chilled, and cut with a compressotome (VF-500–0Z, Precisionary Instruments, Greenville, NC, USA) into 350 μm slices. PCLS were maintained in 24-well plates with Dulbecco’s Modified Eagle Medium (DMEM/F12) (Gibco, Gaithersburg, MD, US), 0.1% human serum, and 1% Antibiotic-Antimycotic (Gibco, Gaithersburg, MD, US) and incubated at 37 °C in 5% CO2 in a humidified chamber.

To evaluate GlucoGlo binding to human tumor, PCLS were incubated with 5µM GlucoGlo in culture media for 1 h, washed with PBS and scanned by the Pearl Trilogy Imaging System. Fresh specimens were formalin-fixed, paraffin-embedded, and sectioned at 5 μm. Fluorescence microscopy was performed on an unstained slide of FFPE tumor with an Odyssey microscopic scanner and Leica DM6 B fluorescence microscope. GlucoGlo fluorescence was correlated with tumor areas on H&E by a board-certified pathologist.

Post-hoc image analysis and statistics

ImageJ (http://rsb.info.nih.gov/ij) was used for post-hoc image analysis and SBR calculations. An SBR > 2 was designated as sufficient fluorescent contrast to delineate tumor from background. Statistical analyses were conducted using GraphPad Prism 8 (GraphPad Software, CA, USA). Fisher’s exact tests and chi-square tests were used for categorical variables where applicable. T-tests and ANOVA were used to compare means, where applicable. P values < 0.05 were considered statistically significant.

Results

GlucoGlo binds SGLT2-expressing cancer cell lines in vitro

We first characterized expression levels of SGLT2 in human NSCLC cell lines (H1299 and H1666) and normal lung fibroblasts (HD28). Immunostaining with anti-SGLT2 antibody revealed high SGLT2 expression in H1299 and H1666 cells, whereas HD28 showed negligible expression (Fig. 1a).

Using SGLT2 expression as a standard, we then measured GlucoGlo binding in the same cell lines by NIR fluorescence microscopy. Consistent with SGLT2 expression patterns, GlucoGlo fluorescence was greatest in the SGLT2-expressing H1299 and H1666 cells (Fig. 1b). In contrast, negligible GlucoGlo fluorescence was observed in the non-SGLT2-expressing HD28 cells, demonstrating that GlucoGlo binds in vitro models of NSCLC in an SGLT2-dependent manner with minimal background fluorescence in non-SGLT2-expressing cells. GlucoGlo binding specificity was confirmed in the high-SGLT2-expressing H1299 cell line, where incubation with excess unconjugated dapagliflozin resulted in negligible fluorescence signal, consistent with competitive inhibition of tracer binding (Supplemental Fig. 2).

Fig. 2.

GlucoGlo accumulates in SGLT2-expressing murine flank xenografts in a dose- and time-dependent manner. (a) Fluorescence overlay of mice with tumors imaged at various times after intravenous delivery of GlucoGlo at increasing doses. (b) Tumor signal-to background ratios (SBRs) plotted over time for each dose. (c) SBRs were compared across doses at various time points, and 0.05 mg/kg was identified as the optimal dose

GlucoGlo accumulates in SGLT2-expressing xenografts

To assess GlucoGlo’s ability to accumulate in SGLT2-expressing tumors in vivo, we evaluated its fluorescence in mice bearing SGLT2-expressing flank xenografts. Mice were administered GlucoGlo via tail vein injection at doses ranging from 0.01 to 0.5 mg/kg and imaged at 3 to 120 h post injection. No adverse events were noted in any mice at any dose.

GlucoGlo demonstrated excellent tumor selectivity following an initial washout phase from surrounding tissue. While early imaging showed background fluorescence across all groups during the first 24 h post-injection, highly selective tumor uptake was observed in the 0.05 and 0.1 mg/kg dosage groups after initial washout (Fig. 2a). At the lowest dose (0.01 mg/kg), the dye initially appeared in the surrounding soft tissue but failed to clearly delineate tumors before washing out. Mice receiving 0.05 mg/kg of GlucoGlo exhibited strong and sustained tumor fluorescence, with mean peak SBRs of 6.85 at 48 h, persisting through 72 h before washing out. Similarly, the 0.1 mg/kg group displayed mean peak SBRs of 8.42 at 72 h (Fig. 2b).

The highest dose of GlucoGlo (0.5 mg/kg) required an extended washout period, with tumor delineation becoming apparent at day 4 post-injection and continuing to increase through the conclusion of the experiment, rendering it less suitable for imaging applications. As peak SBRs were comparable between the 0.05 and 0.1 mg/kg doses (6.85 vs. 8.42, p = 0.32), 0.05 mg/kg was identified as the lowest effective dose (Fig. 2c).

GlucoGlo binding is specific for the SGLT2 transporter

To confirm SGLT2-specific tumor binding, we conducted a competitive inhibition assay using an excess of free SGLT2 inhibitor. Mice pretreated with a 100-fold excess of SGLT2 inhibitor relative to GlucoGlo showed a complete loss of tumor fluorescence with a mean SBR of 1.07, showing that excess SGLT2 inhibitor effectively outcompeted the dye (Fig. 3a). SBRs were significantly greater in the control group compared to the inhibited group (7.55 vs. 1.07 respectively; p < 0.001) (Fig. 3b). These findings confirm the SGLT2-dependent nature of GlucoGlo binding. 

Fig. 3.

GlucoGlo binding is SGLT2-specific. a) Representative images of mice following injection of 0.05 mg/kg GlucoGlo, with 100-fold excess of SGLT2 inhibitor and saline. (b) SBRs were significantly lower in mice pretreated with inhibitor (p < 0.001). Fluorescent areas (c) corresponded with SGLT2 staining on immunohistochemistry (d) which was confirmed as tumor regions on H&E staining (e)

Fluorescent imaging of excised tumor specimens collected 72 h after GlucoGlo administration revealed GlucoGlo signal was distributed throughout the tumor specimen, closely matching the SGLT2 expression pattern observed on IHC (Fig. 3c-d). Microscopic analysis of the H&E-stained sections confirmed that both the fluorescent signal and SGLT2 expression were localized to tumor regions (Fig. 3e), further reinforcing the specificity of GlucoGlo for SGLT2-expressing tumors.

Biodistribution

Ex vivo imaging of organs showed highest GlucoGlo fluorescence in tumor (mean fluorescence: 46.18 a.u.) (Fig. 4a). Minimal or no background fluorescence was observed in the spleen, heart, or brain. The liver, kidneys, and intestine demonstrated moderate background fluorescence (mean fluorescence 29.99, 12.58, and 20.08 a.u., respectively), consistent with the hepatic and renal metabolism of the SGLT2 inhibitor ligand, as well as the physiologic SGLT2 expression in the kidneys [21]. 21% of dapagliflozin is excreted in feces, accounting for the intestinal fluorescence [22]. Mild background fluorescence was also observed in the lungs (mean fluorescence: 12.2 a.u.) and was mainly localized to the larger bronchi. The mean tumor to lung fluorescence ratio was 3.78, indicating ample contrast to support clinical applicability (Fig. 4c).

Fig. 4.

Biodistribution. a) Ex-vivo gross and fluorescence imaging of tumor and organs. Bar graphs demonstrating fluorescence of tumors and organs (b) and the ratios of tumor-to-organ fluorescence (c). (d) Mean body weight of mice measured over time after injection of normal saline control or 0.5 mg/kg GlucoGlo. (e) Percent change in body weight relative to baseline (day 0) was not significantly different between groups

Body weight remained stable in both groups throughout the monitoring period, with no significant differences observed between GlucoGlo-treated and saline control mice in absolute weight (p = 0.98) or percent weight change from baseline (p = 0.62) (Fig. 4d-e). BCS remained unchanged for all animals, and no injection site reactions were observed in either group. No adverse effects were noted grossly in the organs.

GlucoGlo administration improves identification of residual tumor following partial resection

We next investigated whether the addition of IMI with GlucoGlo could improve detection of residual tumor deposits after surgery compared to conventional visualization and palpation techniques. 49 mice received 0.05 mg/kg GlucoGlo and underwent a partial or complete tumor resection (Fig. 5a). The wound beds were assessed by a blinded investigator using visualization and palpation alone and then using IMI.

Fig. 5.

GlucoGlo enhances detection of residual tumor post-resection. a) Representative images of mice after partial and complete tumor resection show GlucoGlo fluorescence in residual tumor deposits. b) Combining conventional visual inspection and palpation with fluorescent GlucoGlo imaging significantly improved residual tumor identification (p < 0.01)

Following pathologic assessment, 22 mice were confirmed as having residual tumor deposits and 27 mice had negative margins. Among the mice with residual tumor, 13/22 were identified by visual inspection and manual palpation, while all 22 residual tumor deposits were found with the addition of IMI. Fluorescent imaging detected a greater proportion of hidden tumor deposits compared to inspection and palpation (100% vs. 59.1%, p < 0.01) (Fig. 5b).

Of the 27 mice with complete resections, 8 mice were incorrectly identified as having positive margins using conventional methods alone. In contrast, IMI correctly identified mice that had no residual tumor in 24 of 27 mice. The addition of fluorescent imaging reduced the incidence of false positive results (11.1% vs. 29.6%, p = 0.18) (Fig. 5c).

Overall, margin assessment with IMI in addition to conventional methods had a sensitivity of 100%, specificity of 88.9%, positive predictive value of 88%, and a negative predictive value of 100%. Conventional methods alone had a sensitivity of 62.5%, specificity of 70.3%, positive predictive value of 61.9%, and negative predictive value of 67.8%.

GlucoGlo accumulates selectively in human LUAD in ex vivo setting

To test the translational potential of the tracer, we performed ex vivo incubation of the tracer with freshly resected patient specimens incubated with GlucoGlo. All patient tumors displayed strong macroscopic fluorescent signal (Fig. 6). Fluorescent microscopy revealed that the tumor areas had significantly higher mean MFI than distant lung tissue (25,218 vs. 3,371; p = 0.009). Mean SBR was 7.57 a.u., indicating excellent tumor-specific labeling of human lung cancer. Of note, the highest SBR was 10.78 a.u. and was confirmed to be a minimally invasive adenocarcinoma on final histology (Fig. 6a). The other tumors were confirmed to be invasive adenocarcinoma.

Fig. 6.

GlucoGlo accumulates selectively in human LUAD. (a) Representative images of human LUAD and distant lung as imaged macroscopically by the Odyssey NIR scanner, and microscopically with H&E and NIR imaging. b) MFI of tumor was significantly higher than distant lung in all cases (p = 0.009), resulting in high signal to background ratios (c)

Discussion

Lung cancer is the leading cause of cancer death in the United States, accounting for nearly 125,000 deaths annually [23]. Surgery remains the cornerstone of curative treatment; however, resections can be challenging due to the difficulty of locating small nodules. As CT surveillance and lung cancer screening have increased, lung cancers are increasingly diagnosed at earlier stages [1, 24]. While this provides an important window for intervention, it introduces new surgical challenges. GGOs and subsolid nodules often resemble benign lung parenchyma and can be nearly impossible to palpate intraoperatively [25]. Furthermore, many patients present with multiple synchronous or metachronous GGOs. As the surgical paradigm shifts from lobectomy to sublobar resections, improved strategies to precisely localize these tumors and assess margins are increasingly necessary to ensure complete resection while preserving healthy lung [26].

In this study, we evaluated GlucoGlo, a NIR probe targeting SGLT2, a marker of early lung cancer transformation, for the intraoperative detection of early-stage NSCLC. GlucoGlo selectively labels SGLT2-expressing NSCLC cells with minimal background fluorescence and demonstrated strong tumor uptake at clinically relevant doses in murine models. Its biodistribution profile and competitive inhibition assays confirmed specificity for SGLT2. Intraoperative imaging with GlucoGlo significantly enhanced margin detection, and GlucoGlo accurately labeled early malignancy in ex vivo human lung tissue, supporting its potential for clinical translation in identifying early cancers during surgery.

Frozen-section analysis is the current standard for intraoperative margin assessment, but it is time-consuming, relies on pathologist expertise, and is subject to sampling error [27]. IMI has shown promise in enhancing intraoperative localization and margin assessment across various solid tumors including ovarian, lung, glioma, and breast cancers [7–10, 28]. Currently, there are no IMI tracers specifically designed to target early-stage lesions. Existing targets, including cathepsins, glucose transporter 1, annexins, and folate receptor, are typically upregulated as malignancies progress, limiting their sensitivity for detecting early malignant transformation [29–34].

SGLT2 is physiologically expressed in the proximal convoluted tubule of the kidney, but has been found to be overexpressed in multiple early-stage cancers, including prostate, pancreatic, breast, and lung cancers and, importantly, it has minimal expression in normal lung tissue [17, 35–38]. Its expression is elevated in preinvasive lung lesions such as atypical adenomatous hyperplasia and adenocarcinoma in situ and is highly expressed in lepidic adenocarcinomas, a common histologic pattern seen in GGOs [14–16]. Because its expression peaks in early lung cancer, SGLT2 is well suited as a marker for early disease. Our study demonstrated that GlucoGlo, an SGLT2-targeted NIR probe, selectively binds to SGLT2-expressing cells both in vitro and in vivo. It exhibited negligible staining of non-malignant lung fibroblasts and showed excellent tumor to background ratios in mice, demonstrating strong specificity for the SGLT2 target. Furthermore, when assessing surgical margins, the use of GlucoGlo IMI significantly improved the surgeon’s sensitivity for detecting positive margins as small as 1 mm without increasing the rate of false positives. Moreover, our data suggested a trend toward improved identification of negative margins when GlucoGlo was used as an adjunct to traditional visualization and palpation, countering concerns that IMI agents might increase false positives from background fluorescence.

Several findings support GlucoGlo’s potential for clinical translation. In murine models, GlucoGlo achieved excellent tumor delineation at clinically translatable doses and maintained sustained high SBRs through 72 h, suggesting a flexible window of time points at which a patient could be imaged in a clinical setting. It had no obvious toxic effects in mice, as expected given its composition of the FDA-approved SGLT2 inhibitor dapagliflozin with demonstrated safety in both patients with and without diabetes [39, 40]. Furthermore, GlucoGlo utilizes an ICG fluorescent moiety, which offers improved depth of penetration and reduced autofluorescence compared to dyes in the visible light spectrum [20, 41]. As ICG is compatible with most clinically available NIR imaging devices, GlucoGlo is well positioned for streamlined clinical translation.

While this study demonstrates significant potential for GlucoGlo as an intraoperative imaging agent, we acknowledge some limitations. Our findings are currently limited to preclinical and ex-vivo models. Ultimately, human clinical trials will be needed to characterize the benefits and limitations of the dye. The sample size for our ex-vivo PCLS investigations is small and lacks benign controls. Further study is needed regarding the specificity for the dye in differentiating infectious or inflammatory lung lesions, particularly those presenting as ground glass opacities. Although existing literature suggests SGLT2 is not constitutively upregulated in inflammatory or fibrotic tissue [42, 43], direct comparative evaluation in human inflammatory lung specimens was not performed here and warrants further investigation. Additionally, although the adjunct use of IMI with GlucoGlo significantly improved sensitivity for positive margins, its ability to improve specificity for negative margins was less definitive. This may be attributed to the limitations of our flank xenograft model, which lacks the complexity and variability of lung tissue. Future studies using orthotopic lung models may better evaluate GlucoGlo’s potential to rule out malignancy in benign lung parenchyma.

Conclusion

GlucoGlo is a promising intraoperative imaging agent for resection of early lung cancers. By targeting SGLT2, a biomarker expressed early in NSCLC development, GlucoGlo addresses a key gap in current imaging strategies. Its strong specificity, low background fluorescence, and compatibility with standard NIR imaging devices make it an ideal candidate for clinical translation. With further evaluation in human studies, GlucoGlo could significantly enhance surgical precision, improve margin detection, and ultimately improve outcomes for patients undergoing lung cancer resection.

Supplementary Material

Below is the link to the electronic supplementary material.

Author contributions

KOW, KAM, JT and SS contributed to study conception and design. Material preparation, data collection and analysis were performed by KOW, KAM, JT, XS, PZ, VB, and SS. The first draft of the manuscript was written by KOW and all authors edited and approved the final manuscript.

Funding

KOW was supported by the surgical oncology training grant T32-CA251063-05 from the National Institutes of Health. SS and KAM were supported by grant P01 CA254859 from the National Institutes of Health. The content of the manuscript is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Data availability

The datasets generated during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The Animal Care and Use Committee of the University of Pennsylvania approved all animal study protocols, and experiments were conducted in compliance with the Guide for the Care and Use of Laboratory Animals. Included patients provided informed consent for tissue donation. The study was performed in line with the principles of the Declaration of Helsinki and was approved by the University of Pennsylvania Institutional Review Board. 

Consent for publication

Not applicable.

Competing interests

The authors have no other relevant financial or non-financial interests to disclose.