Intraoperative Molecular Imaging in Lung Cancer: The State of the Art and the Future

Main Text

Lung cancer is the deadliest malignant disease in the United States and has a median survival of only 5–11 months at advanced stages.¹ However, for patients with stage I–IIIA cancers, minimally invasive surgery with curative intent plays a key role in treatment,² and improving signal to noise in identifying diseased tissue is therefore critical in guiding resection. Non-targeted dyes, such as fluorescein (FITC) or indocyanine green (ICG), have long been administered pre-operatively to improve tumor contrast and have significantly improved surgical outcomes.3, 4 More recently, targeted probes have been developed that utilize antibodies, aptamers, or peptides to bind to cancerous tissue more specifically and further enhance the tumor-to-background ratio (TBR).5, 6 In this issue of Molecular Therapy, Predina et al.⁷ report on their pilot clinical trial of a near-infrared small-molecule probe targeting folate receptor-α (FRα) to sensitively detect pulmonary adenocarcinomas (PAs) during surgery. In line with previous clinical trials of FRα-targeted probes in ovarian and colorectal cancers and of fluorescently-labeled anti-EGFR antibodies, such as cetuximab-IRD800, in head and neck cancer, this trial lends further support to the clinical utility of intraoperative molecular imaging (IMI) in identifying lesions that might otherwise be missed.8, 9, 10, 11

Over the last few years, the research group of Sunil Singhal has studied several IMI agents to optimize resection of PAs. In an early study, 18 patients who presented with a pulmonary nodule (PN) requiring surgical removal were administered 5 mg/kg ICG intravenously. ICG is a US Food and Drug Administration (FDA)-approved near infrared (NIR) fluorophore that is not target specific on its own, but it is retained in tumors by the enhanced permeability and retention (EPR) effect.¹² Using this method, surgeons identified 16 out of 18 PNs, as well as 5 additional nodules that were not identified by standard preoperative imaging, at a depth of up to 1.3 cm from the pleural surface.¹³

In order to improve the specificity of IMI in lung cancer, focus then shifted to testing targeted probes.¹⁴ With the discovery of a high prevalence of FRα expression in two major non-small cell lung cancer (NSCLC) types–80% of PAs and 20%–40% of squamous cell carcinomas (SCC)–FRα has become an attractive molecular target in NSCLC. Two clinical trials using the probe EC17 (folate conjugated to FITC) were conducted by the Singhal group. In one study, surgeons identified 19/19 PAs by comparing ex situ fluorescence with histological findings from 30 PN biopsies.¹⁵ In a second trial of 50 patients with biopsy-confirmed PA who underwent surgery with EC17, 92% of tumors fluoresced during surgery with a mean TBR of 4.2 ± 2.7. Additionally, previously unidentified nodules were found in two patients, leading to re-staging of their cancers.¹⁶ A later murine study by the group demonstrated that EC17 could also be employed to identify residual tumor cells at surgical margins.¹⁷ Despite the improved target specificity of EC17, however, untargeted ICG was more successful in identifying new nodules (Table 1) simply because of the greater tissue penetration possible in the NIR range compared with visible-range dyes.

Table 1.

Clinical Trials of Different Target-Fluorophore Combinations in IMI of Pulmonary Adenocarcinoma

Study	Target	Fluorophore	No. of Patients	Sensitivity (%)	TBR	Depth (mm)	% of Patients with Additional Nodules Found	Minimum Tumor Size Detected (mm)
Okusanya et al., 201413	none	ICG	18	89	2.2	13	28	2
Okusanya et al., 201516	FR-α	FITC	50	92	4.2	1	4	3
Predina et al., 20177	FR-α	SO456	10	100	3	<30	30	12

Note the increased penetration depth and greater fraction of patients in whom additional nodules were found using NIR agents.

To overcome this limitation, a natural next step was to study an FRα-targeted probe in which FITC was replaced with an NIR dye. OTL38 is a probe consisting of folate conjugated to S0456, a dye with a similar chemical structure to NIR cyanine dyes, including IR783 and Cy7, which have higher quantum yields (8.4%–28%) than ICG (7.8%).¹⁸ A major advantage of small molecule probes, like OTL38 (∼1.4 kDa), over bulkier antibody-based probes is rapid plasma clearance, while still maintaining a long residence time in tumors, allowing administration shortly before surgery and providing the surgeon with a long detection window.⁷ By contrast, antibody-based probes (∼150 kDa) must be administered days in advance for adequate washout time in order to obtain low background levels. While OTL38 is currently being tested in clinical trials of IMI in ovarian and renal cell carcinoma, the study by Predina et al.⁷ is among the first to demonstrate its use in PAs. The investigators found that OTL38 was detectable deeper in tissue than FITC (1 cm versus 1 mm) and that tissue autofluorescence was considerably lower in the NIR range. OTL38 demonstrated superior TBR compared with visible-range dyes (FITC, PPIX) as well as ICG at all depths tested (down to 10 mm below the pleural surface). Moreover, the authors showed that NSCLC cells with higher FRα expression showed correspondingly higher binding of OTL38. In orthotopic mouse xenografts implanted with A549, an NSCLC cell line highly expressing folate, nodules with a mean size of 2.4 ± 0.7 mm were detected with a mean TBR of 2.9 ± 0.6.

The investigators then sought to translate the probe clinically. Upon review of clinicopathologic data from 100 pulmonary resection specimens, FRα expression was observed in 86% of PA specimens. In a subsequent clinical study, 10 patients with PNs were administered 0.025 mg/kg OTL38 at 3–6 hours before surgery. 8 patients had nodules that could be detected intraoperatively by fluorescence through the pleura (TBR ∼3), all of which proved to be FRα-positive PAs. The non-fluorescent nodules were a squamous cell carcinoma and a pulmonary hamartoma (both FRα negative). In two subjects, PNs were found on preoperative imaging that had no positron emission tomography (PET) avidity but displayed strong fluorescence intraoperatively and turned out to indeed be PAs expressing FRα. These results are promising, and a much larger cohort, including patients with non-PA malignancies, will be needed to accurately assess the sensitivity and specificity of OTL38 in PA and NSCLC.

The study by Predina et al.⁷ reported a number of particularly exciting findings that merit further discussion by the authors. First, in three subjects, synchronous disease was detected by IMI that was not detected by preoperative imaging or intraoperative visualization and palpation. In one subject, this changed the operative plan and, in two cases, the patients were upstaged from stage IA to stage III, thus changing their postoperative treatment. These results are consistent with a larger IMI study on PA by the same group and previous findings on OTL38 in ovarian cancer.8, 19 However, given that detection of lesions <5 mm by computed tomography (CT) is common in the lung,²⁰ one wonders why these 6–8 mm nodules were missed in preoperative imaging. Second, the authors state that they could not reliably detect lesions deeper than 3 cm, implying that lesions at depths of up to 3 cm were detectable. If reproducible, this result is truly intriguing, given that most IMI studies with NIR probes have not detected lesions deeper than ∼1 cm. Third, while the authors observed a TBR of 3, comfortably above the commonly reported threshold TBR (∼2) for reliable tumor identification, it is noteworthy that, in a study of OTL38 in ovarian cancer, a TBR of 4.4 was required to clearly detect tumors.⁸ Finally, the authors observed no adverse events from OTL38 in this small cohort, but others have previously described dose-dependent mild-to-moderate hypersensitivity reactions in 30 healthy volunteers⁸ and in ∼18% of ovarian cancer patients given the same 0.025 mg/kg dose of OTL38 in a separate phase II trial. For clinical translation of OTL38, rigorous toxicity testing beyond assessment for acute toxicity is warranted, including assessment of histological changes in organs and tissues over an extended period of time post-administration in preclinical animal models. In addition, prospective studies with large numbers of patients will be necessary to assess the impact of IMI with OTL38 on progression-free survival and overall survival. Nevertheless, OTL38 appears to overcome many of the shortcomings of past IMI probes due to its specificity, imaging depth, TBR, and pharmacokinetics, making it a highly promising candidate for clinical translation.

IMI is a robust surgical tool with great potential to improve tumor resection, outcomes, and survival. The state of the art in IMI utilizes probes consisting of a highly targeted binder linked to a NIR fluorophore, and the development of an increasing number of such agents for molecular targets in a wide variety of cancers is anticipated. There are considerable challenges to overcome, however, including limitations on the depth at which tumors can be detected and the number of targets that can be probed simultaneously. In this regard, other optical agents and modalities, used alone or in combination, could further advance the field. For example, detection at depths of several centimeters is possible by using dyes that fluoresce in the NIR-II region (1,000–1,700 nm).²¹ These dyes also enable higher spatial resolution—1 μm at 1–3 mm depth compared with 0.2 mm depth for standard NIR-I dyes (e.g., ICG, IRDye800)—due to reduced background in the NIR-II region. However, using NIR-II dyes would require special camera systems, detectors, and lasers that are uncommon at present, limiting their widespread use.²² Raman imaging could allow for far greater multiplexing than fluorescence, as spectra from over 10 different Raman dyes, each linked to a different binder, could be read out simultaneously. This could also enable in vivo proteomic diagnostics. Moreover, background white light in the operating room or in endoscopy does not interfere with Raman imaging. Raman signals can be enhanced using gold core/silica shell nanoparticles; however, none have been approved thus far for human use.²³ Other modalities, including photoacoustics and radiofrequency (RF)-acoustics, can allow for much deeper detection in tissue than Raman or fluorescence approaches but are limited by spatial resolution. Combined use of photoacoustics and fluorescence has been proposed, wherein photoacoustics, with its greater detection depth, would guide the surgeon to the area of interest, and fluorescence would then offer the resolution needed for resection of the tumor margins.²⁴

Radioguided surgery (RGS) has been used in the clinic for decades and employs a hand-held probe for intraoperative detection of injected radionuclides. Gamma probes are commonly used with single photon emission computed tomography (SPECT) radionuclides, like Technetium-99m (^99mTc), to identify sentinel lymph nodes and occult lesions, an established procedure in breast cancer management. Despite numerous promising studies, however, the use of ^99mTc colloids in the surgical management of lung cancer has not become common practice and targeted strategies using SPECT-labeled antibodies in lung adenocarcinoma have proven problematic due to the high radioactive background from the heart and blood vessels.²⁵ PET imaging, like SPECT, is commonly used with CT and is widely used in oncology for perioperative evaluation.26, 27 PET is valued not only for its sensitivity and depth of detection but also its ability to yield quantitative information. Despite the frequent use of PET-CT in pre-operative assessment of solid tumors like NSCLC, true intra-operative PET scanning has not been possible due to the size and geometric constraints of scanners.28, 29 Portable hand-held PET probes for intraoperative detection of high energy γ and β rays have also been developed and evaluated in a limited number of clinical studies over the past decade.30, 31, 32, 33 However, these probes have encountered limited success due to current probe design and the sensitivity and specificity of the radiotracer (¹⁸Fluorodeoxyglucose) itself. Due to their synergistic properties, preoperative PET and intraoperative fluorescence are already used in conjunction.¹⁹ However, dual-labeled PET-NIR contrast agents, which have yet to be developed, could enable co-registration of PET and fluorescence signals and could therefore serve an important role in future clinical management.³²

While the intraoperative use of other molecular imaging modalities is still in its infancy, fluorescence image-guided surgery is developing a track record of feasibility, safety, and utility and is rapidly becoming the state of the art in surgical navigation. Several clinical trials are underway for a number of cancer types, and completed trials have thus far reported positive outcomes overall. Future technologies will afford improvements in detection depth and signal to noise and will likely harness artificial intelligence to assist surgeons or even serve as completely autonomous supervised substitutes.³⁴