If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Address for reprints: Sunil Singhal, MD, Department of Surgery, University of Pennsylvania Perelman School of Medicine, 3400 Civic Center Blvd, 14th Floor, South Pavilion, Philadelphia, PA 19104.
Intraoperative molecular imaging (IMI) using tumor-targeted optical contrast agents can improve thoracic cancer resections. There are no large-scale studies to guide surgeons in patient selection or imaging agent choice. Here, we report our institutional experience with IMI for lung and pleural tumor resection in 500 patients over a decade.
Methods
Between December 2011 and November 2021, patients with lung or pleural nodules undergoing resection were preoperatively infused with 1 of 4 optical contrast tracers: EC17, TumorGlow, pafolacianine, or SGM-101. Then, during resection, IMI was used to identify pulmonary nodules, confirm margins, and identify synchronous lesions. We retrospectively reviewed patient demographic data, lesion diagnoses, and IMI tumor-to-background ratios (TBRs).
Results
Five hundred patients underwent resection of 677 lesions. We found that there were 4 types of clinical utility of IMI: detection of positive margins (n = 32, 6.4% of patients), identification of residual disease after resection (n = 37, 7.4%), detection of synchronous cancers not predicted on preoperative imaging (n = 26, 5.2%), and minimally invasive localization of nonpalpable lesions (n = 101 lesions, 14.9%). Pafolacianine was most effective for adenocarcinoma-spectrum malignancies (mean TBR, 2.84), and TumorGlow was most effective for metastatic disease and mesothelioma (TBR, 3.1). False-negative fluorescence was primarily seen in mucinous adenocarcinomas (mean TBR, 1.8), heavy smokers (>30 pack years; TBR, 1.9), and tumors greater than 2.0 cm from the pleural surface (TBR, 1.3).
Conclusions
IMI may be effective in improving resection of lung and pleural tumors. The choice of IMI tracer should vary by the surgical indication and the primary clinical challenge.
Local recurrence after resection complicates the treatment course of a substantial portion of patients with thoracic tumors. Intraoperative molecular imaging (IMI) is a technology that has been developed as a surgical adjunct to improve resection completeness. Here, we review our experience with IMI-guided resection of lung and pleural tumors, which spans 500 patients and 4 different IMI tracers.
The optimal therapy for lung cancer, selected cancers metastatic to the lungs, and pleural cancers is complete surgical resection. However, many patients suffer locoregional recurrence after curative-intent resection due to a limitation in surgeons' ability to identify and completely remove disease using visual inspection and palpation alone.
Therefore, there is an unmet need to develop technologies to serve as adjuncts to clinical judgment in detecting occult lung and pleural cancer deposits during surgery.
To aid in detection of occult malignancy, intraoperative molecular imaging (IMI) has been developed to complement traditional surgical methods. This technology involves the systemic administration of a tumor-targeted fluorescent contrast agent in the preoperative period.
The tracer then accumulates in malignant tissues and can be detected by wavelength-specific cameras to allow real-time visual identification of lesions in patients and in the specimen on the back table (Figure 1).
Figure 1Overview of intraoperative molecular imaging. A, Patients are preoperatively infused with a targeted optical contrast agent that localizes to cancer cells. During surgery, a wavelength-specific thoracoscope is used to identify the fluorescent tracer that has localized to malignant cells. Resected specimens can be imaged on the back table using a wavelength-specific exoscope to interrogate the resection margins. B, Timelines of “eras” in IMI-guided pulmonary resection. Initial efforts focused on non-NIR receptor targeted tracers, followed by NIR tracers (both nonreceptor-targeted and receptor-targeted). Future work in IMI should be guided toward addressing limitations of the technology in its current form. Activity-based probes can improve tracer specificity and reduce false-positive fluorescence. NIR-II probes can improve depth of signal penetration, whereas the administration of cocktails of receptor-targeted tracers can expand the range of tumors identifiable by IMI. Finally, optical imaging approaches such as NIR-CLE can be used to move from macroscopic tumor imaging to microscopic identification of cancer cells during biopsy procedures. NIR, Near infrared; CLE, confocal laser endomicroscopy; ICG, indocyanine green.
Over the last decade, our group has tested 4 tumor-specific fluorophores in the visible (380-700 nm) and near-infrared (NIR, 700-1000 nm) light spectra to guide resection of lung cancer, mesothelioma, and other cancers of the chest and mediastinum.
These include nonreceptor-targeted agents such as high-dose (5 mg/kg) indocyanine green (ICG; TumorGlow), which accumulates in sites of increased vascular permeability such as the tumor microenvironment via the enhanced permeability and retention effect.
Other imaging agents that have been studied are targeted to cell-surface receptors that are overexpressed on primary lung cancers and metastases to the lung, such as folate receptor alpha (FRα) and carcinoembryonic antigen (CEA).
Although the technique is being applied to a wide range of tumor types, there are no large-scale studies to guide surgeons in selecting an imaging agent or understanding the patient population for whom the technology will be the most clinically useful.
To address this knowledge gap, we retrospectively reviewed our decade-long experience with IMI-guided resection of lung and pleural tumors, which spans 500 patients and 4 different IMI tracers. To our knowledge, this is the largest experience in the world. Although previous publications from our group have explored the strengths and limitations of individual IMI tracers, none have directly compared imaging agents for different indications in thoracic surgery. Our objective in this study was to investigate the strengths and limitations of IMI across tracer types to provide a practical guide for surgeons in patient and tracer selection as the technology gains wider adoption.
Methods
Summary of Study Design
Between December 2011 and November 2021, 500 patients with lung and pleural nodules suspicious for malignancy were enrolled in clinical trials of intraoperative molecular imaging. All patients provided written informed consent before enrollment, and all studies were approved by the University of Pennsylvania Institutional Review Board (approval numbers 815,058, 811,870, 820,766, 822,153, 826,547, 831,268, 834,577, 842,586, 844,554, 850,299, and 850,893). All trials were registered at ClinicalTrials.gov (EC17: NCT01778920; TumorGlow: NCT02280954, NCT02621268, NCT01335893; pafolacianine: NCT02602119, NCT02872701, NCT04241315; SGM-101: NCT04315467). Inclusion criteria were patients older than 18 years of age with a primary diagnosis or a high clinical suspicion of lung nodules warranting surgery based on computed tomography and/or positron emission tomography imaging. Exclusion criteria were any medical condition that in the opinion of the investigators could potentially jeopardize the safety of the patient, pregnancy, history of anaphylactic reactions, impaired renal or hepatic function, or known sensitivity to fluorescent light. For this review, we included all patients who had ever undergone an IMI-guided resection of a lung or pleural nodule at the University of Pennsylvania. All patient data were prospectively entered into a database, and clinically significant events were predetermined and noted in real time.
Depending on the study drug used and cancer type, patients were infused with the respective imaging agents between 3 hours and 4 days before surgery. Intraoperatively, surgeons used thoracoscopic visualization and finger palpation to identify known lesions. Next, an optical imaging device with settings optimized to detect the relevant imaging agent was used to capture signal from the fluorescent probe. This device was also used to inspect the lung for additional nodules. After resection, specimens were imaged ex vivo and margins were assessed by IMI as described to follow. Specimens were then submitted for pathologic analysis by a board-certified thoracic pathologist.
Study Drugs
Folate-fluorescein (EC17; C42H36N10O10S; molecular weight: 872.87 Da) is a folate analog conjugated to fluorescein isothiocyanate (see Figure E1 for molecular structure of all study drugs). Fluorescein isothiocyanate is a synthetic compound that excites at a wavelength of 465 to 490 nm and emits in the 520 to 530 nm range. A 0.1 mg/kg dose of EC17 was dissolved in 10 mL of saline and administered intravenously 4 hours before surgery. EC17 was provided by On Target Laboratories. EC17 targets the FRα, which is overexpressed in 90% of pulmonary adenocarcinomas.
ICG (C43H47N2NaO6S2; molecular weight: 775 Da) is a nonreceptor-targeted NIR contrast agent with excitation and emission wavelengths of 805 nm and 830 nm, respectively. TumorGlow is a high-dose formulation of ICG (5 mg/kg) that has a propensity to accumulate in a range of tumor types via the enhanced permeability and retention effect.
Patients were infused with 1 to 5 mg/kg of the drug intravenously over 40 minutes at various time points before surgery. ICG was provided by Akorn Pharmaceuticals.
Pafolacianine (OTL38; C61H63N9Na4O17S4; molecular weight: 1414.42 Da) is a conjugate between folate and the NIR fluorescent dye S0456. Pafolacianine excites at 774 to 776 nm and emits at 794 to 796 nm. A 0.025 mg/kg dose of the drug was diluted in 20 to 220 mL saline or 5% dextrose and infused intravenously over 60 minutes 3 to 24 hours before surgery. The tracer targets FRα.
Pafolacianine was obtained from On Target Laboratories.
Anti-CEACAM5–conjugated fluorochrome (SGM-101, C61H63N9Na4O17S4; molecular weight: 1414.42 Da) is a conjugate between SGM-ch511, an anti-CEACAM5 chimeric monoclonal antibody, and the NIR fluorochrome BM-104. SGM-101 excites at 685 nm and emits at 705 nm. SGM-101 targets CEA receptor, which is overexpressed in a subset of primary pulmonary adenocarcinomas as well as lung metastases from colorectal cancers.
Patients received a 10-mg dose of the drug intravenously over 30 minutes 4 days before surgery. SGM-101 was provided by SurgiMAb.
Imaging Devices
All imaging devices were optimized for detection of the specific wavelength emitted by the individual tracers. EC17 imaging was performed using the Flocam (BioVision Technologies Inc), as described previously.
Pafolacianine imaging was performed using the Quest Artemis (Quest Medical Imaging), Medtronic Visionsense3 (Medtronic), Stryker AIM-C1, or DaVinci Firefly (Intuitive Surgical) as previously described.
Back-table analysis was performed using a Medtronic VisionSense3 exoscope, a LI-COR ELVIS prototype (LI-COR Biosciences), or the LI-COR Pearl Trilogy imaging system (LI-COR Biosciences). SGM-101 imaging was performed using the Quest Artemis and Quest Spectrum (Quest Medical Imaging).
Histopathologic and Fluorescent Microscopic Review of Specimens
Excised specimens were formalin-fixed and paraffin-embedded. Sequential 5-μm sections were obtained and underwent comprehensive histopathologic and fluorescent analysis. All specimens were reviewed by a board-certified thoracic pathologist. Unstained 5-μm sections were evaluated using a NIR microscopic scanner (Odyssey; LI-COR Biosciences) and a NIR microscope (Leica Microsystems).
Analysis of Clinically Significant Events Relating to IMI
To understand the clinical value of IMI, we reviewed all cases for clinically significant events (CSEs). We have previously coined this term as a benchmark for evaluating clinical trials of IMI and use it to identify any occurrence related to IMI that caused the surgeon to modify the operation, identify additional cancers, or upstage the patient's cancer.
In our retrospective review, we identified 4 principal CSEs related to IMI in our large series of 500 patients. Three were related to in situ IMI findings, and one involves back-table applications of IMI.
Statistical Analysis
Post hoc image analysis was conducted using ImageJ (Image J; https://imagej.nih.gov/ij/). Mean fluorescence intensity of lesions and background lung tissue was obtained by measuring the region of interest in monochromatic NIR images. Background fluorescence was obtained from tissue 0.5 to 1.0 cm from the margin. A minimum of 1000 pixels were included in each measurement. Calculations were repeated in triplicate, and tumor-to-background ratios (TBRs) were calculated with a value greater than 2.0 considered fluorescent. All comparisons were made using Stata Statistical Software: Release 14 (StataCorp LP). The Student t test was used unless otherwise indicated.
Results
Patient and Lesion Characteristics
Over the course of our decade-long experience, 500 patients underwent surgery to remove a total of 677 lung and pleural lesions. Mean patient age was 63.8 years (interquartile range [IQR], 59-72 years). The majority of patients were female (n = 297, 59.4%) and former smokers (n = 272 [54.4%], 29.0 mean pack years). Patients underwent infusion with either pafolacianine (n = 271, 54.2%), TumorGlow (n = 154, 30.8%), EC17 (n = 71, 14.2%), or SGM-101 (n = 4, 0.8%). All study drugs were well tolerated, and there were no serious adverse events related to tracer infusion. Mean nodule size was 2.34 cm (IQR, 1.10-2.90 cm), and mean depth was 0.6 cm from the pleural surface. By histopathologic analysis, the majority of resected lesions were primary lung cancers (n = 411, 60.7%), followed by metastatic lesions (n = 108, 16.0%), mesothelioma (n = 82, 12.1%), and benign lesions (n = 76, 11.2%). Within the primary lung cancer subgroup, most lesions were adenocarcinoma-spectrum lesions (n = 332, 80.8% of primary lung cancers in the study). A full summary of patient and lesion characteristics is provided in Table 1.
Table 1Patient and lesion characteristics
Patient characteristics
Number (%) or mean [IQR]
Total patients
500
Age
63.8 [59-72]
Sex
Male
203 (40.6%)
Female
297 (59.4%)
Race
White
413 (82.6%)
Black
51 (10.2%)
Asian
16 (1%)
Other/unknown
28 (5.6%)
Former smokers
272 (54.4%)
Pack years
29 [21-40]
Tracer
EC17
71 (14.2%)
ICG
154 (30.8%)
Pafolacianine
271 (54.2%)
SGM-101
4 (0.8%)
Lesion characteristics
Total lesions
677
Size of lesion, cm
2.34 [1.1-2.9]
PET SUV
4.9 [1.8-6.7]
Tumor location
Lung
582 (86.0%)
RUL
152 (22.5%)
RML
61 (9.0%)
RLL
147 (21.7%)
LUL
123 (18.2%)
LLL
99 (14.6%)
Pleura
83 (12.1%)
Chest wall
12 (1.7%)
Pathologic diagnosis
Primary lung cancer
411 (60.7%)
Invasive adenocarcinoma
286 (42.2%)
Minimally invasive adenocarcinoma
26 (3.8%)
Adenocarcinoma in situ
20 (3.0%)
Squamous cell carcinoma
45 (6.6%)
Small cell lung cancer
8 (1.1%)
Carcinoid tumor
26 (3.8%)
Mesothelioma
82 (12.1%)
Metastatic lesion
108 (16.0%)
Benign lesion
76 (11.2%)
Tumor differentiation
Well differentiated
113 (16.7%)
Moderately differentiated
199 (29.4%)
Poorly differentiated
125 (18.5%)
Not reported
240 (35.6%)
IQR, Interquartile range; ICG, indocyanine green; PET, position emission tomography; SUV, standardized uptake value; RUL, right upper lobe; RML, right middle lobe; RLL, right lower lobe; LUL, left upper lobe; LLL, left lower lobe.
In our review, one of the most important benefits provided by IMI is NIR imaging of the specimen once it is removed. Specifically, we found that we could accurately predict tumor-positive margins by the presence of fluorescence on the staple line (Figure 2, A). In total, IMI was able to detect close (<5 mm) or tumor-positive margins in 32 (6.4%) patients. By imaging resected specimens with a ruler for scale, we found that we could quantify the resection margin distance using IMI. IMI images were analyzed by measuring the distance of a centimeter in pixels and then measuring from the edge of the fluorescent signal to the staple line. IMI margin distance was compared with the margin distance reported on final pathology. We applied the technique to 79 specimens, and found that IMI margins were nearly identical to margins reported on final pathology (P = .65, R2 = 0.872, Figure 2, A). The median difference in margins was 2.4 mm (IQR, 0.9-3.6 mm).
Figure 2Representative examples of clinically significant events relating to IMI. A, IMI-guided margin evaluation. Back-table IMI identified tumor-positive margins that were confirmed on H&E and fluorescence microscopy. Furthermore, margin distance calculation by IMI was highly accurate when compared with formal pathology margin calculation (R2 = 0.872). B, Preoperative and intraoperative images of a patient undergoing pleurectomy for mesothelioma. IMI identified residual disease after resection that was not apparent on brightfield imaging. C, Preoperative and intraoperative images of a patient undergoing resection of a metastatic osteosarcoma. Two lesions were identified on preoperative imaging, but IMI detected an additional 2 lesions not seen on preoperative CT or PET. D, Preoperative and intraoperative images of a patient undergoing resection of a subpleural ground-glass opacity that was not palpable or visually apparent by pleuroparenchymal distortion. IMI clearly identified the lesion. IMI, Intraoperative molecular imaging; H&E, hematoxylin and eosin; CT, computed tomography; PET, position emission tomography.
IMI Can Identify Residual Disease After Cytoreductive Surgery
The second CSE involves identification of residual disease after resection (see Figure 2, B, for a representative case). This CSE occurred in 37 patients (7.4%) in our series, all of whom were undergoing cytoreductive or debulking surgery for mesothelioma. These patients underwent IMI with high-dose ICG (TumorGlow), which was able to identify residual deposits of disease as small as 1 mm. The mean TBR of residual disease deposits was not significantly different than the TBR of disease identified by conventional techniques in the initial resection (3.4 vs 2.9, P = .21). We found that residual lesions were frequently identified in anatomic locations that are difficult to visualize, such as the costophrenic sulcus or directly adjacent to thoracotomy site. In addition, residual disease was likely to be identified in areas of significant pleural stripping and where acute inflammation and blood shielded them from traditional visual inspection.
Identification of Synchronous Malignancies
The third CSE relates to identification of synchronous cancers not predicted on preoperative imaging or visualized by the surgeon (see Figure 2, C, for a representative case). Synchronous nodules were identified in 26 (5.2%) of the 500 subjects. The vast majority of these nodules (n = 25, 96.1%) that were positive (ie, fluorescent) by IMI were malignant by histopathology. The majority of synchronous malignancies (n = 20, 76.9%) were identified using TumorGlow during pulmonary metastasectomy for sarcoma and colorectal metastases. We found that the average size of synchronous nodules identified by IMI alone was 0.7 cm (IQR, 0.4-1.1 cm). This was significantly smaller than the nodule that the surgeon was seeking, which measured 2.5 cm on average (IQR, 2.1-2.8 cm, P < .0001). Despite differences in size, mean TBR of synchronous nodules was similar to those pulmonary nodules identified preoperatively (2.4 vs 2.9, P = .45).
Localization of Nonpalpable Lesions During Minimally Invasive Resection
Another major added value of IMI involved localizing lesions that the surgeon was unable to locate using visual inspection and finger palpation. In total, 101 lesions (14.9%) in our study were not localized by traditional means of palpation or visual inspection but were localized by IMI. All of these patients were undergoing video-assisted thoracoscopic surgery or robotic-assisted thoracic surgery. Lesion localization by IMI spared these patients the morbidity of a thoracotomy or an extended resection such as a segmentectomy or lobectomy, which would have been the alternative if traditional methods to localize the tumors failed. In subgroup analysis, this CSE most commonly occurred during resection for subpleural ground-glass opacity lesions (GGOs). The average size of these lesions was 1.3 cm (IQR, 1.1-1.6 cm).
Optimal Tracer Selection and Dosing Varies by Lesion Histology
In analyzing our series, we found important differences in the efficacy of IMI tracers for different types of thoracic cancers. With regard to TumorGlow, we found that the technique was most effective in identifying metastatic lesions from nonpulmonary malignancies as well as mesothelioma. However, the optimal dose of TumorGlow varied between primary non–small cell lung cancer (NSCLC) as compared with other malignancies (OMs). We found that low-dose TumorGlow (1-3 mg/kg) was less effective for primary NSCLC but could identify OM, the majority of which were pulmonary metastases from distant primary malignancies (Figure 3, A, mean TBR 3.04 vs 1.59, P < .0001). Only 16.2% of NSCLCs were fluorescent when imaged after administration of low dose ICG as compared with 84.6% of OMs. There were no significant differences in TBR when high-dose ICG (4-5 mg/kg) was used for resection of both OM and NSCLC (TBR = 3.73 vs 3.01, P = .06).
Figure 3Differences in tracer performance by indication for operation. A, Violin plot of TBRs for NSCLC compared with OMs when stratified by ICG dosing level (low dose: 1-3 mg/kg, high dose: 4-5 mg/kg). Mean TBR for OM was significantly greater than NSCLC for low-dose ICG, but there was no difference in TBR at high-dose ICG. Graph at right shows the percentage of lesions fluorescent at low-dose versus high-dose ICG. ∗∗P < .01. B, Representative H&E and fluorescence microscopy imaging showing OM imaged with low-dose ICG compared with NSCLC imaged with high-dose ICG. The red circles indicate the location of the malignant lesion and illustrate the pseudocapsule commonly seen in OM, which helps retain ICG compared with the more diffuse spread of NSCLC. Scale bars represent 5 mm. C, Representative examples of IMI with pafolacianine and ICG for subpleural GGOs. D, Plot of TBRs of subpleural GGOs imaged with ICG and pafolacianine. Mean TBR was significantly greater for pafolacianine. Means with standard deviations are plotted with each individual point representing a single lesion. Graph at right shows the percentage of GGOs fluorescent when imaged with ICG versus pafolacianine. ∗∗P < .01. ICG, Indocyanine green; NSCLC, non–small cell lung cancers; IMI, intraoperative molecular imaging; H&E, hematoxylin and eosin.
To explain the mechanism behind the dose-dependence of labeling by TumorGlow we examined resection specimens after staining with hematoxylin and eosin and compared them with tissue sections examined under fluorescence microscopy. We found that metastatic lesions were often well circumscribed with an outer pseudocapsule allowing for accumulation and retention of ICG (Figure 3, B, for representative lesions). Primary NSCLCs lacked this pseudocapsule, resulting in greater diffusion of the fluorochrome into the surrounding lung parenchyma.
When we restricted our analysis to subpleural adenocarcinoma-spectrum malignancies, we found that pafolacianine-based IMI had greater efficacy than alternative approaches (see Figure 3, C, for representative cases). Mean TBR of subpleural GGOs imaged with pafolacianine was 2.84, compared with 1.87 for TumorGlow (Figure 3, D, P = .001). Eleven of 24 (45.8%) of subpleural GGOs were visible by TumorGlow as compared with 25 of 33 (75.8%) by pafolacianine-guided IMI. The majority of GGOs (n = 62, 88.6%) in our series were adenocarcinoma-spectrum lesions, which often overexpress FRα and can therefore be targeted by pafolacianine and EC17. However, because of the visible spectrum of EC17, the depth of penetration into the lung parenchyma was severely limited.
Mechanisms of Failure of IMI: False-Negative Fluorescence
One of the most important values we obtained from the review of our institutional series was a better understanding of the sources of false-negative fluorescence. Of the 601 malignant lesions included in the study, 477 (79.3%) were fluorescent by IMI. In the remaining 124 (21.7%) malignant lesions, we investigated various patient and lesion factors associated with reduced fluorescence. Among lesion factors, we first examined the effect of lesion depth on fluorescent signal. In all tracers, we found that the mean fluorescence intensity and TBR decreased as a function of lesion depth from the pleural surface and frequency of fluorophore emission wavelength (Figure 4, A). Depth of lesion detection was significantly greater for NIR tracers (TumorGlow and pafolacianine) than for the visible spectrum tracer EC17 (1.7 vs 0.3 cm, P < .001). This trend was consistent across all histologic subtypes of lung cancer. Additional lesion-related causes of false-negative fluorescence were mucinous adenocarcinoma histologic subtype (Figure 4, B, n = 19, mean TBR 1.8) and absence of FRα expression in lesions undergoing IMI with pafolacianine (n = 30, mean TBR, 1.2).
Figure 4Mechanisms of false-negative fluorescence in IMI. A, Scatter plot showing the negative correlation between depth of lesion from the pleural surface and tumor-to-background ratio. Each dot represents a single lesion and different colors indicate the tracer used. NIR tracers (pafolacianine and ICG) had much greater depth of signal penetration than the non-NIR tracer (EC17). B, Representative intraoperative images of various cases of false-negative fluorescence in IMI-guided pulmonary resection. ICG, Indocyanine green; IMI, intraoperative molecular imaging; H&E, hematoxylin and eosin.
Regarding patient factors associated with false-negative fluorescence, we found that smoking history was the predominant patient-related predictor of false-negative fluorescence irrespective of lesion depth or histologic subtype. Specifically, we found that patients with more than a 30 pack-year smoking history had mean TBR of 1.9 as compared with 2.6 for lesions in patients with a lesser smoking history and never smokers (P < .001). This reduced TBR was primarily driven by the accumulation of light absorbing carbons (LACs) related to cigarette smoking (see Figure 4, B, for a representative case).
Technical factors related to false negative fluorescence included posterior tumors due to the practical limitations in visualizing nodules in these locations. We further found that use of a 30° thoracoscope provided inferior fluorescence imaging as compared to a 0o thoracoscope due to tissue scattering effects and the optical physics of capturing emitted light from the fluorophores.
Mechanisms of Failure of IMI: False-Positive Fluorescence
Along with understanding the mechanisms of false-negative fluorescence, we also sought to understand the mechanisms of false positivity in IMI. Among the 76 benign lesions in the study, we found that 26 (34.2%) were fluorescent by IMI. By histopathological analysis, all false positives were benign inflammatory lesions, including granulomas (Figure 5, A, n = 13), inflammatory myofibroblastic lesions (Figure 5, B, n = 3), reactive lymphoid hyperplasia (n = 3), subpleural fibrosis (n = 3), foreign body giant cell reaction (n = 2), and organizing pneumonia (n = 2). Mean TBR of these lesions was 5.1 (IQR, 2.1-7.2). Previous work by our group has shown that for pafolacianine-guided IMI, the tracer is taken up by FRβ-expressing macrophages in false-positive lesions.
Figure 5Mechanisms of false-positive fluorescence in IMI. A, Representative preoperative and intraoperative images showing white light and IMI images of a fluorescent granuloma. B, Representative preoperative and intraoperative images showing white light and IMI images of a fluorescent inflammatory myofibroblastic tumor. CT, Computed tomography; PET, position emission tomography; IMI, intraoperative molecular imaging.
In this study, we have reviewed our decade-long institutional experience with IMI-guided resection of pulmonary and pleural cancers. To our knowledge, this is the most extensive experience with IMI in thoracic surgery, and examination of our patient series has uncovered several key findings. First, there are 4 major clinical benefits of IMI in thoracic surgery: interrogating resection margin adequacy, identifying synchronous lesions not detected on preoperative imaging, highlighting residual disease after resection, and localizing lesions that are undetectable by palpation or pleuroparenchymal distortion. The second key finding of our study is a comprehensive elucidation of the sources of false-positive and false-negative fluorescence in IMI-guided thoracic surgery. Third, this study provides important context for surgeons in patient and tracer selection as the technology moves toward wider clinical adoption.
Our work builds on a large body of literature studying IMI in thoracic surgery. Initial experiences in IMI were centered on receptor-targeted fluorescent tracers in the far-red spectrum.
Subsequent work in ICG-guided IMI demonstrated the benefits of using a fluorophore with NIR fluorescence to partially mitigate the effect of lesion depth on optical signal.
We identified several clinical pearls in selecting patients who are most likely to benefit from the technology. First, the choice of IMI tracer should vary by indication for operation (Table 2). Although adenocarcinoma-spectrum lesions represent the majority of the NSCLC lesions in our study, we do believe that the tracers are useful and effective in identifying malignancies of other histologic subtypes. For instance, pafolacianine may be the most helpful for identifying subpleural adenocarcinomas, as many express FRα. A particular subset that benefits is GGOs, which tend to have soft-tissue architecture and can be exceedingly difficult to distinguish from normal lung parenchyma. SGM-101 is likely to provide greatest clinical utility in the identification of lung metastases from a primary colon cancer, given the propensity of these lesions to express CEA. TumorGlow may be most effective for identification of synchronous lesions in patients undergoing pulmonary metastasectomy or detection of residual disease in those undergoing mesothelioma resection. The second clinical pearl in our series is that patients with deep lesions (>2 cm from the pleural surface) or a heavy smoking history (>30 pack years) are less likely to benefit from IMI due to depth of signal penetration. We do not advocate that this technology should be used for all pulmonary resections. Indeed, there are certain clinical scenarios in which the technology may not have an impact on the scope or conduct of the operation (eg, a large, biopsy-proven primary lung cancer that would require lobectomy). We find the greatest utility for the technology in minimally invasive resections, where the surgeon's primary means of interacting with the surgical field is visual information relating to the location of the lesion. This is particularly true in robotic surgery, where there is no haptic feedback to guide the surgeon. However, we find that the technology has utility in open resection as well, particularly in the case of margin evaluation.
Table 2Suggested applications for IMI tracers
Indication
Clinical challenge
Suggested tracer
Mechanism of action
Limitations
Subpleural GGO
Localization Margin evaluation
Pafolacianine
FR-targeted
1) Limited efficacy in lesions lacking FRα expression 2) False-positive fluorescence in benign lesions due to macrophages expressing FRβ 3) Anthracotic lungs 4) Mucinous adenocarcinomas
Adenocarcinoma
Localization Margin evaluation
Pafolacianine
FR-targeted
1) Limited efficacy in lesions lacking FRα expression 2) False-positive fluorescence in benign lesions due to macrophages expressing FRβ 3) Anthracotic lungs 4) Mucinous adenocarcinomas
Sarcoma lung metastasis
Detection of synchronous lesions
TumorGlow
EPR effect
1) Anthracotic lungs 2) False-positive fluorescence due to passive nodal drainage from lesion
Colorectal lung metastasis
Detection of synchronous lesions
SGM-101
CEA-targeted
1) Limited efficacy in lesions lacking CEA expression 2) Limited depth of penetration
This study has better elucidated important limitations in IMI, which provide direction for future research efforts in the field. First, we have shown that false-negative fluorescence is primarily due to lack of signal penetration in deep lesions or the lack of target receptor expression. To address the issue of signal penetration, future work in tracer development should focus on fluorophores in the NIR-II spectrum (>1000 nm), which will be able to highlight deeper tumors.
To address the issue of target receptor expression, ongoing research should examine the ability of cocktails of different tracers (each targeting a different receptor) to expand the range of tumors identified by IMI.
Although our study has demonstrated important clinical benefits of IMI in improving surgical outcomes, the impact of this technology on patient survival needs to be studied. Preliminary work from our group found a possible survival benefit for ICG-based IMI in sarcoma metastasectomy,
but these findings need to be confirmed in large, prospective trials and evaluated in other tumor types.
Another important limitation of IMI is false-positive fluorescence. In our experience, there is a learning curve to IMI interpretation in real time, particularly when distinguishing tracer signal from background noise in the nonreceptor-targeted agents. To address this issue, future investigation should focus on developing greater specificity in tracers. Promising work in this direction has been conducted by several groups investigating activity-based probes, which are tracers that become fluorescent after processing by a tumor-specific enzyme or in response to a physiologic change common to many tumors.
Finally, existing efforts in IMI have primarily concerned macroscopic tumor imaging during surgery, but the next frontier of the technology will be identifying microscopic, single-cell quantities of disease using optical approaches such as NIR confocal laser endomicroscopy.
As this technology gains wider clinical adoption, important considerations regarding health care costs and resource use need to be addressed. The initial capital outlays for initiating an IMI program are significant, requiring the purchase of specialized imaging equipment, training personnel, and hiring staff to coordinate additional patient visits. However, the potential cost-savings stemming from clinically significant events as a result of IMI may provide net cost benefits. For instance, if IMI is able to lower the rate of positive margins, there may be net savings in health care costs. A recent analysis estimated that the total health care cost of a positive lung cancer resection margin is between $60,000 and $129,000. In addition, if IMI is able to reduce the number of conversions to thoracotomy for lesion localization, there may be significant cost savings due to reduced operative time and shortened length of hospital stays for patients. These benefits should be important considerations as hospitals and healthcare systems consider initiating IMI programs.
In summary, this study has analyzed our experience with 500 patients undergoing resection of lung cancer or mesothelioma guided by IMI with 4 different optical tracers. We found that IMI has clinical utility in a number of specific aspects of thoracic surgery, and we also identified patient subgroups that may be less likely to benefit from this technology. As the use of IMI continues to expand, future work should be focused on addressing the limitations of the technology in its present form.
Conflict of Interest Statement
The authors reported no conflicts of interest.
The Journal policy requires editors and reviewers to disclose conflicts of interest and to decline handling or reviewing manuscripts for which they may have a conflict of interest. The editors and reviewers of this article have no conflicts of interest.
Appendix E1
Figure E1Overview of IMI tracers. This figure demonstrates the optical properties of all tracers investigated at our institution, detailing their optical properties, molecular structures, and representative images of their use. ICG, Indocyanine green; EPR, enhanced permeability and retention; FR, folate receptor; CEA, carcinoembryonic antigen; H&E, hematoxylin and eosin.
Dr Kennedy was supported by the American Philosophical Society and the National Institutes of Health (grant F32 CA254210-01). Dr Azari was supported by the Society for Thoracic Surgeons and the Stephen C.C. Leung Fellowship. Dr Singhal was supported by the National Institutes of Health (grant P01 CA254859).
00093-4&id=fx1.jpg){kind=link}
00093-4&id=gr1.jpg){kind=link}
00093-4&id=gr2.jpg){kind=link}
00093-4&id=gr3.jpg){kind=link}
00093-4&id=gr4.jpg){kind=link}
00093-4&id=gr5.jpg){kind=link}
00093-4&id=fx2.jpg){kind=link}