Prospective validation of tumor folate receptor expression density with the association of pafolacianine fluorescence during intraoperative molecular imaging–guided lung cancer resections

Abstract

Purpose

Pafolacianine, a folate receptor alpha-targeted NIR tracer, has demonstrated clear efficacy in intraoperative molecular imaging–guided (IMI) lung cancer surgery. However, the selection of patients who would benefit from IMI remains challenging given the variability of fluorescence with patient-associated and histopathologic factors. Our goal in this study was to prospectively evaluate whether preoperative FRα/FRβ staining can predict pafolacianine-based fluorescence during real-time lung cancer resections.

Methods

This was a prospective study conducted between 2018 and 2022 that reviewed core biopsy and intraoperative data from patients with suspected lung cancer. A total of 196 patients were deemed eligible, of whom core biopsies were taken from 38 patients and assessed for FRα and FRβ expression by immunohistochemistry (IHC). All patients underwent infusion of pafolacianine 24 h prior to surgery. Intraoperative fluorescence images were captured with the VisionSense bandpass filter–enabled camera. All histopathologic assessments were performed by a board-certified thoracic pathologist.

Results

Of the 38 patients, 5 (13.1%) were found to have benign lesions (necrotizing granulomatous inflammation, lymphoid aggregates) and 1 had metastatic non-lung nodule. Thirty (81.5%) had malignant lesions, with the vast majority (23, 77.4%) being lung adenocarcinoma (7 (22.5%) SCC). None of the benign tumors (0/5, 0%) exhibited in vivo fluorescence (mean TBR of 1.72), while 95% of the malignant tumors fluoresced (mean TBR of 3.11 ± 0.31) compared to squamous cell carcinoma (1.89 ± 0.29) of the lung and sarcomatous lung metastasis (2.32 ± 0.09) (p < 0.01). The TBR was significantly higher in the malignant tumors (p = 0.009). The median FRα and FRβ staining intensities were both 1.5 for benign tumors, while the FRα and FRβ staining intensities were 3 and 2 for malignant tumors, respectively. Increased FRα expression was significantly associated with the presence of fluorescence (p = 0.01),

Conclusion

This prospective study sought to determine whether preoperative FRα and FRβ expression on core biopsy IHC correlates with intraoperative fluorescence during pafolacianine-guided surgery. These results, although of small sample size, including limited non-adenocarcinoma cohort, suggest that performing FRα IHC on preoperative core biopsies of adenocarcinomas as compared to squamous cell carcinomas could provide low-cost, clinically useful information for optimal patient selection which should be further explored in advanced clinical trials.

Introduction

Lung cancer is currently the second most prevalent cancer and the leading cause of cancer-related deaths globally [1, 2]. As such, advancement in the treatment of lung cancer is a priority. Although noninvasive approaches such as chemotherapy, radiation, and immunotherapy have seen significant advances, surgery remains a mainstay in the treatment of lung cancer [3]. Nevertheless, the success of surgical cure rests on achieving negative margins, with studies repeatedly showing that negative and wide margins correlate with a lower risk of recurrence as well as improved overall survival [4, 5]. However, with the advent of minimally invasive resection options, the current surgical armamentarium relies on centuries-old visual inspection and tactile feedback techniques, making the detection of tumor borders difficult, particularly in cases of nonpalpable ground glass lesions [6, 7]. Intraoperative molecular imaging (IMI) has emerged to address these challenges, using targeted fluorophores to clearly localize tumors in real time with minimal impact on patient safety and surgical workflow [8].

Non-small cell lung cancer (NSCLC), particularly adenocarcinomas, has a high expression of plasma membrane glycoprotein folate receptor alpha (FRα), which binds to serum folate 10³–10⁴ times more strongly than normal lung epithelium, making FRα an attractive candidate for targeted IMI [9]. Pafolacianine, a cyanine-based near-infrared (NIR) dye conjugated to folic acid, is a recently developed contrast agent designed to target FRα [7, 10]. It thereby enables the fluorescent localization of lung adenocarcinomas [11]. Pafolacianine has demonstrated strong results in phase I and II trials; in the phase I trial, all (21/21) preoperatively identified lung adenocarcinomas displayed fluorescence, and in the phase II trial with 110 patients, pafolacianine located nine cancerous lesions that were not detected preoperatively and improved outcome for 26% of patients [12, 13]. A phase III ELUCIDATE trial (NCT04241315) is ongoing to follow up on these encouraging results with data expected to be available over the coming months. The drug is currently under review for the management of lung malignancies by the FDA.

Although pafolacianine was designed to target FRα, cross-reactivity with folate receptor beta (FRβ) exists due to the 71% amino acid homology shared between FRα and FRβ [9, 14-16]. Despite this homology, FRα and FRβ maintain distinct cellular distributions [14, 17]. While FRα is expressed in both normal and cancerous lung epithelium, FRβ expression is restricted to immune cells, such as monocytes and tumor-associated macrophages [14]. These expression patterns provide unique strengths for pafolacianine-guided IMI, as lung adenocarcinomas contain a mix of epithelial cells and tumor-associated macrophages [16]. For example, pafolacianine fluorescence may derive from both the epithelial and immune components of lung cancers, contributing to increased sensitivity for detecting adenocarcinomas.

The ability of pafolacianine to target both FRα and FRβ also provides potential guidance on fluorochrome selection for IMI-guided lung surgeries. The selection of patients who would benefit from IMI is challenging given the variability of fluorescence with patient-associated factors, such as cancer histology, tumor depth, presence of mucin, and anthracosis from light-absorbing carbons [15, 18, 19]. This is paramount when assessing for FRβ expression in nodules with significant tumor necrosis as this is associated with abundance of tumor-associated macrophages which express FRβ, which can cause false positivity. Additionally, lungs with significant anthracosis have elevated inflammatory markers leading to high FRβ, leading to potential false positivity [15, 20].

With preoperative biopsies increasingly being taken in the management of lung cancer to assess the feasibility of immune checkpoint inhibitor use, there is now an opportunity to additionally assess tumor FRα and FRβ expression with little additional cost. Strong expression of either marker may favor pafolacianine-guided surgeries, while low expression would instead prompt usage of alternative fluorophores or increase the surgeon’s suspicion of nonfluorescent yet palpable lesions. Additionally, optimal selection of patients can prevent rare infusion-related clinical side effects and optimize healthcare costs by recognizing which patients would benefit from pafolacianine-guided lung nodule resections.

In this study, we prospectively reviewed core biopsy data from patients with lung cancer, and we hypothesized that preoperative FR expression correlates with intraoperative fluorescence. Biopsies were assessed for FRα and FRβ expression by immunohistochemistry (IHC) and correlated with diagnosis by pathology. We further evaluated the concordance between biopsies to determine the minimum number of samples necessary to guide fluorochrome selection. Our results suggest that strong expression of FRα and FRβ predicts intraoperative fluorescence with high accuracy but that multiple core biopsies need to be taken to avoid false negative results. To our knowledge, this is the first study that prospectively analyzed histopathologic factors as a predictive factor for intraoperative pafolacianine-based fluorescent labeling of lung malignancies and can serve as a useful adjunct in future IMI patient selection.

Materials and methods

Study design and patient selection

The current study was an open-label, proof of principle study performed at the Hospital of the University of Pennsylvania in the USA. The study was reviewed and approved by the University of Pennsylvania Institutional Review Board in accordance with Good Clinical Practice (GCP) guidelines as outlined by the international council of Harmonization of Technical Requirements for Pharmaceuticals for Human Use in addition to laws and regulations for human research by the state of Pennsylvania and Federal Drug Authority (FDA). Preoperative analysis was performed as a subgroup of the NCT04241315 (ELUCIDATE) trial.

Overall inclusion criteria

Patients aged 18 years or older with suspicion of primary lung malignancy who were scheduled for anatomic segmentectomy/lobectomy/pneumonectomy via video-assisted thoracoscopic resection (VATS), thoracotomy, or median sternotomy were included in the study. Patients were screened and evaluated for pulmonary nodules with cross-sectional imaging, including fine-cut 1-mm thickness respiratory-gated CT scanning and PET-CT. Patients with multivisceral disseminated disease who would not benefit from curative lung resection were excluded from the study. A negative serum pregnancy test at screening followed by a negative urine pregnancy test on the day of surgery or day of admission for female patients of childbearing potential, female patients of childbearing potential or less than 2 years postmenopausal agrees to use an acceptable form of contraception from the time of signing informed consent until 30 days after study completion, ability to understand the requirements of the study, provide written informed consent and authorization of use and disclosure of protected health information, and agree to abide by the study restrictions and to return for the required assessments.

Overall exclusion criteria

Exclusion criteria for enrollment in the studies included previous exposure to fluorescent tracers, any medical condition that in the opinion of the investigators could potentially jeopardize the safety of the patient, history of anaphylactic reactions or severe allergies, history of allergy to any of the components of folic acid or pafolacianine, pregnancy, positive pregnancy test, clinically significant abnormalities on electrocardiogram (ECG) at screening, presence of any psychological, familial, sociological, or geographical condition potentially hampering compliance with the study protocol and follow-up schedule, impaired renal function defined as estimated glomerular filtration rate (eGFR) < 50 mL/min/1.73 m², impaired liver function, and severely altered hematologic parameters including thrombocytopenia, neutropenia, and transfusion requiring anemia. Additional non-laboratory exclusion criteria included patients who received an investigational agent in another investigational drug or vaccine trial within 30 days prior to surgery, had known sensitivity to fluorescent light, underwent chemoradiotherapy within 6 weeks of surgical intervention, had a Karnofsky performance status less than 70%, had a median ASA class 4 or higher, had other preexisting terminal diagnoses, and had an ejection function lower than 30%.

Specimen analysis

Patients with peripheral or peribronchial lesions amenable for preoperative core biopsies of suspicious nodules scheduled for surgical curative resection were selected for this cohort. Patients were prospectively followed, and the immunohistochemical expression pattern correlation with intraoperative fluorescence was analyzed. Core biopsies of the samples were then processed by an experienced dedicated thoracic pathologist. Tissues were formalin fixed and paraffin embedded for FR staining and immunofluorescence. The technique for immunohistochemical analysis for folate receptor expression has been previously described and includes five-micron sections of formalin-fixed, paraffin-embedded tissue stained using antibodies against anti-human folate receptor alpha (clone BN3.2, Leica NCL-L-FRαlpha, 1:40 dilution) and folate receptor beta (Novus NBP2-43,654, 1:500 dilution). Staining was performed on a Leica Bond III instrument using the Bond Polymer Refine Detection System (Leica Biosystems DS9800) according to the manufacturer’s instructions. Heat-induced epitope retrieval was performed for 20 min in Epitope Retrieval Solution 2 (Leica Biosystems AR9640). Folate receptor alpha required extended primary antibody incubation (60 min.).

NIR tracer

Analyzed patients underwent infusion of the study drug pafolacianine (0.025 mg/kg) up to 24 h prior to surgery. Pafolacianine (C₆₁H₆₃N₉Na₄O₁₇S₄, MW: 1414.42) is a folate analog conjugated to the NIR dye S0456. Pafolacianine excites at 774–776 nm and emits at 794–796 nm. Specifically, pafolacianine targets folate receptor alpha and folate receptor beta, which are expressed in more than 85% of adenocarcinomas of the lung [9].

Surgical procedure

Surgical procedures were performed by a thoracic surgeon. During surgery, surgeons utilized standard visualization and finger palpation (when applicable) to identify known tumors. After identification of the tumor, NIR imaging was used to confirm lesion fluorescence. If the preoperatively identified nodule was unidentifiable by white-light visualization or palpation, localization using fluorescence guidance was attempted. After identifying the primary lesions, fluorescence imaging was utilized to assess the hemithorax for occult lesions. Following resection of the nodules, the specimen was analyzed ex vivo on a back table using an NIR camera system to assess nodule fluorescence and margin assessment. The sample was then sent for frozen section analysis by a board-certified thoracic pathologist.

Camera systems

In situ fluorescent imaging was performed using a Vision-Sense Iridium (Medtronic, Minneapolis, MN) imaging system. VisionSense Iridium IMI-based optical devices are high-definition, dual-band (white light and NIR) camera systems capable of concurrent NIR emission and detection. The tracer-specific iridium system utilizes an excitation laser with a wavelength of 780–805 nm, with fluorescence detection based on a bandpass filter selective to light ranging from 790 to 850 nm. VisionSense Iridium uses two charge-coupled device sensors to produce simultaneous white-light RGB and NIR fluorescence images from the emission channel and merges the two in real time. It is coupled to a dedicated 4-mm outer diameter endoscope and exoscope attachment that allows for a wider visualization area. Both endoscope and exoscope attachments have a dual optical path design that enables the separate use of white light (visible light) and NIR light [7].

The mean fluorescence intensity (MFI) of the tumors and normal lung parenchyma were calculated from ImageJ (National Institutes of Health) and MATLAB (MathWorks, Natick, MA) software with a minimum of 1000 pixels included. The tumor-to-background ratio (TBR) was calculated for the specimens.

Statistical analysis

Data are presented as the mean (standard deviation) unless otherwise noted. Data were analyzed for parametric distribution. Parametric correlation of dichotomous variables was compared using Student’s t test and ANOVA for 3 or more continuous variables. Nonparametric categorical variables between groups were compared using the Kruskal–Wallis test, whereas the Mann–Whitney U test was conducted for continuous variables. Statistical analyses were performed using R version 3.5.3 and the packages tidyverse and survminer (R Foundation for Statistical Computing, Vienna, Austria). p values<0.05 were considered statistically significant. Additionally, both parametric and nonparametric statistical analyses were performed using SPSS version 27 (IBM Technologies).

Results

Demographics and study population

We aimed to enroll a consecutive group of patients who were deemed high risk on cross-sectional imaging or based on smoking history who would require tissue biopsy for confirmation and subsequently selected for curative surgery. Given that the goal of our study was to evaluate immunohistochemical markers prospectively based on preoperative histopathologic assessment, only patients with tissue diagnosis were selected for final analysis. Between 2018 and 2022, 196 patients underwent pafolacianine-guided surgical resection. Of these, 38 patients with preoperative tissue sampling were included in this study. Twenty-six patients were prospectively followed and analyzed for tumor biomarkers and their association with pafolacianine-guided intraoperative fluorescence and tumor localization. The remaining patients were either lost to follow-up or did not follow standard post-infusion clinical follow-up. We selected 7 patients for internal quality control (5 malignant, 2 benign) to determine the number of core biopsies/tissue sampling required for accurate folate receptor analysis and its association with pafolacianine-labeled tumor fluorescence. Two patients were lost to follow-up, and 2 could not undergo thoracoscopic IMI-guided resection; these patients were excluded from the final analysis.

Overall, the majority of patients were female (n = 24, 63.2%), white (n = 25, 65.7%), and had a significant smoking history (29.1 PPY). The average lesion size included in the study was 1.21 cm. Preoperative FDG PET/CT was SUV_max 3.76. Most malignant lesions in the study (77.4%) were adenocarcinoma lesions, with squamous cell carcinoma comprising quarter of the malignant nodules.

All the patients received 0.025 mg/kg of pafolacianine up to 24 h prior to surgical resection with a median time of 19.12 h from infusion to IMI-guided nodule removal. There were no pafolacianine-associated complications and no 30- or 90-day mortality noted in the study. The median American Society of Anesthesiologist (ASA) physical classification was 3 for all comers in the study. Details of demographics and preoperative characteristics are displayed in Table 1.

Table 1.

Demographics table of the patients analyzed in the study, including histopathologic characteristics on postoperative analysis

	Number (%) and/or median [IQR]
Patient characteristics
Total patients enrolled	38/196
Study arm patient (n)
Folate receptor preop IHC score	26 (68%)
Core biopsy internal quality control	7(18%)
Patients lost to follow-up	3 (10.1%)
Received pafolacianine infusion without surgery	2 (5%)
Age (years)	63.68 [51.1–70.21]
Sex
Male	14 (36.8%)
Female	24 (63.2%)
Race
White	25 (65.7%)
Black	6 (15.7%)
Asian	3 (7.8%)
Other/unknown	3 (7.8%)
Any smoking history	32 (84.2%)
Pack years	29.1 [10.5–42.3]
Time from infusion to resection (h)	19.12 [16–23.45]
Pafolacianine dose (mg/kg)	0.025
Median ASA classification	3
Hospital length of stay (days)	1.32 (1.11–3.2)
30-day mortality	0
90-day mortality	0
Lesion characteristics
Total malignant primary lung lesions analyzed	30
Size of lesion (cm)	1.21 [0.71–2.4]
PET SUV	3.76 [1.1–6.03]
Tumor location
RUL	9 (30.0%)
RML	8 (25.8%)
RLL	7 (22.5%)
LUL	3 (9.7%)
LLL	3 (9.7%)
Pleura	0
Chest wall	0
Final pathology
Invasive adenocarcinoma	23 (77.4%)
Squamous cell carcinoma (SCC)	7 (22.5%)
Small cell lung cancer	0
Metastasis	1 (3.2%)
Benign lesion	5 (16.1%)
Tumor differentiation
Well differentiated	19 (63.3%)
Moderately differentiated	3 (9.6%)
Poorly differentiated	8 (25.8%)

Preoperative folate receptor immunohistochemistry on biopsy specimens

Next, we wanted to explore whether the preoperative diagnosis of FR + IHC correlated with pafolacianine NIR labeling during IMI-guided resections. Forty-seven biopsies were taken in the preoperative setting. A total of 28/47 were performed via transthoracic needle aspiration (TTNA), and 19/47 were performed using endobronchial ultrasound bronchoscopy. A total of 31/47 of the lesions analyzed were diagnosed as lesions potentially harboring malignancy by board-certified thoracic pathologists.

Immunohistochemical analysis of the preoperative samples demonstrated that out of 30 lesions, 22/30 had FRα detected, with 16/22 having at least 2 +scoring. A total of 12/30 lesions had FRβ detected, with 4/12 having a strong (3 +) FRβ scoring.

A total of 29/30 malignant lesions demonstrated NIR fluorescence either in situ, ex vivo, or on tumor bisection. The two patients who did not demonstrate any fluorescence had highly necrotic tumors as well as anthracosis. However, on fluorescence microscopy, both patients demonstrated NIR fluorescence at an emission wavelength of 800 nm. Patients with lung adenocarcinoma had the highest tumor-to-back-ground ratio (TBR) at 3.11 ± 0.31 compared to squamous cell carcinoma (1.89 ± 0.29) of the lung and sarcomatous lung metastasis (2.32 ± 0.09) (p < 0.01) (Fig. 1A-D).

Fig. 1.

Malignant lesions demonstrate pafolacianine labeling. A Patient with a diagnosis of adenocarcinoma on preoperative biopsy demonstrating pafolacianine-guided in situ localization during IMI-guided resection. B Fluorescence microscopy analysis demonstrates NIR localization of pafolacianine to areas of adenocarcinoma on (inset) with C areas of the tumor demonstrating uptake on pafolacianine at 63 × magnification. D Adenocarcinoma lesions demonstrate the highest tumor-to-background ratio (TBR) compared to squamous cell carcinoma of the lung and lung metastases, but malignant lesions overall have higher fluorescence intensity than benign lesions. E The preoperative FDG PET standard uptake value did not correlate with malignant lesion fluorescence

Histopathologic and fluorescence microscopy analysis of the malignant tumors demonstrated tumor-specific labeling by pafolacianine (Fig. 1B, C). Areas of nodules with malignant cells demonstrated increased NIR fluorescence emission with a median mean fluorescence intensity (MFI) of 314 ± 65 absorbance units (A.U), whereas there was minimal NIR emission detection in areas with normal lung parenchyma or no tumor (54 ± 9 A.U.) (p < 0.05). This was evident even at the single-cell level at 63 × magnification, demonstrating high tumor targeting by pafolacianine.

Additionally, given the correlative observation of association with malignancy and NIR emission of tumor after pafolacianine labeling, we wanted to explore if preoperative FDG PET imaging would be predictive of intraoperative fluorescence as it is widely performed in majority of patients with suspected lung cancer. Patients with a diagnosis of cancer were grouped into two groups: SUV < 2.5 and SUV > 2.5 as measured by a dedicated thoracic radiologist. Subgroup analysis demonstrated no predictive ability or correlation with intraoperative fluorescence by pafolacianine in our small cohort (Fig. 1E) (p = 0.43).

Human folate receptor plasma membrane glycoprotein density associated with pafolacianine fluorescence

Due to the high fluorescence emission intensity of pafolacianine in adenocarcinomas, we wanted to explore whether preoperative immunohistochemical analysis would be predictive of which patients would benefit from pafolacianine-guided lung cancer resections. Pafolacianine is a folate receptor alpha (FRα)–targeted NIR fluorochrome, whereas adenocarcinomas of the lung have been demonstrated to express FRα > 85% of the time. The association of FRα staining with successful IMI localization has been demonstrated on postoperative analysis, but preoperative FRα scores have not been validated prospectively.

In our cohort of patients, the presence of FRα overall was strongly associated with the detection of in vivo and ex vivo fluorescence of pafolacianine (p < 0.05) (Fig. 2A, D). Interestingly, although there was a trend for increased intensity in the fluorescence emission of pafolacianine with a higher density of FRα expression, as there was tumor surface presence of FRα, pafolacianine could be detected intraoperatively (p < 0.05) (Fig. 2B, D). The FRα score correlated with higher TBR scores, with an R² of 0.71 (Fig. 2C). Tumors (squamous cell carcinoma) that did not fluoresce intraoperatively or were found to have minimal to no expression of FRα on preoperative IHC analysis (Fig. 3). This is consistent with a prior analysis of SCC tumors that only expressed FRα less than 33% of the time.

Fig. 2.

FRα presence is associated with pafolacianine NIR labeling intraop. A Analysis of all lesions demonstrates that any FRα presence is associated with statistically higher fluorescence versus minimal to no fluorescence when there is no FRα expression by the tumor (p < 0.05). B There was a correlation of a higher tumor-to-background ratio with increased expression of FRα, but this correlation did not reach statistical significance. Furthermore, there was no difference in fluorescence intensity between 2 + and 3 + FRα scoring. C There is a strong correlation with FRα presence and increased fluorescence when analyzing patients with adenocarcinoma diagnosis on preoperative sampling. D Representative images of a patient with adenocarcinoma diagnosed on core biopsy with a 3 + FRα score by a pathologist and observation of pafolacianine NIR labeling

Fig. 3.

Squamous cell cancer lesion with negative FRα staining and lack of fluorescence during pafolacianine-guided resection

Folate receptor β weakly correlated with NIR fluorescence:

One of the confounders with pafolacianine-guided resections has been NIR labeling of granulomatous lesions, benign lymph nodes, and lymphoid aggregates. Recently, this has been postulated to be due to the presence of tumor-associated and activated macrophages, which have been demonstrated to express FRβ [8, 19]. However, the degree of FRβ density and NIR emission intensity has not been fully elucidated in a prospective manner. We analyzed the preoperative samples for FRβ IHC staining and aimed to analyze its predictive ability for fluorescence detection during IMI-guided resections using pafolacianine.

Initially, we confirmed FRβ as the predominant source of pafolacianine fluorescence emission in benign lymph nodes sampled from 10 consecutive patients. None of the lymph nodes was found to harbor malignancy, but 8/10 demonstrated fluorescence on ex vivo analysis. Only FRβ presence along with CD 68 (marker for tumor-associated macrophages) was observed in these lymph nodes, and this correlated with areas of signal emission on fluorescence microscopy (p < 0.05). Representative findings are shown in Supplementary Fig. 1.

We then explored whether this remained true for malignant lesions (Fig. 4). We examined lesions where FRα expression was low and FRβ was present. In 4 patients where there was no FRα but FRβ was detected, only the lesions with a 3 + score (2 patients) had fluorescence detected. The 1 + and 2 + lesions did not demonstrate statistically significant fluorescence, with TBR values below 1.7 (p < 0.05). The presence of FRβ other than the strong 3 + that is present in lymph nodes is the only weakly correlated factor for pafolacianine fluorescence. This can be seen in a representative image (Fig. 4C, bottom) from a patient with a lymphoid aggregate where there is minimal FRα expression but strong FRβ with fluorescence noted on IMI. However, (Fig. 4C, top) lack of FRβ was not correlated with lack of IMI detection. This remained true on statistical analysis where FRβ expression did not reach statistical significance with TBR or fluorescence presence in general (p = 0.49).

Fig. 4.

FRβ weakly correlated with NIR fluorescence. A Regardless of the fluorescence intensity that was observed, there was no correlation with FRβ presence (B), with an R² of only 0.02. B In the subgroup analysis, only strong FRβ expression (3 +) was noted to be associated with TBR > 2. C Representative images of malignant and benign lesions showing FRβ not correlating with detection or lack of detection during IMI

Is one tissue sample enough?

Due to observation that folate receptor alpha is strongly predictive for pafolacianine fluorescence emission detection during IMI-guided resection, we wanted to explore to see if there is concordance with different core biopsies obtained from the same tumor. Select patients (n = 7) who were deemed safe to undergo IMI-guided lung nodule resection were selected for this analysis. These patients (nodule size > 1.5 cm) underwent 5 core biopsies of the tumor, and each sample was then stained for FRα and FRβ to determine if all the IHC staining was concordant (Fig. 5).

Fig. 5.

Representative core biopsy samples from various patients with different pathologies. Patient 3 (bottom) and sample 5 did not stain for FRα, whereas the previous 4 biopsies had a strong presence of FRα on IHC

Of all the malignant lesions (n = 5), 4/5 patients had concordant findings in all 5 biopsies for both folate receptor alpha and beta. One of the patients with the diagnosis of adenocarcinoma had one core biopsy not stained for folate receptor alpha, whereas the previous 4 biopsies were strongly positive for FRα presence (Fig. 5). There was variable scoring for FRβ. IHC staining with a kappa score for interobserver agreement of 0.67, where the kappa score was 0.91 for FRα when the samples were analyzed by 9 different independent clinicians, including pathologists (Fig. 5).

Discussion

Intraoperative molecular imaging (IMI) is emerging as a powerful tool in the armamentarium of surgical oncologists [21-23]. The technology allows real-time intraoperative localization of nonpalpable visually occult disease, assessment of margin safety, and identification of synchronous lesions missed on extensive preoperative workup. As such, hundreds of new tracer developments are currently being studied in various clinical trials. Pafolacianine (cytalux), as a folate receptor–targeted NIR nonionizing fluorescent tracer, has been extensively studied over the last decade as an IMI agent in thoracic oncology with encouraging results. However, preoperatively selecting or excluding patients for pafolacianine-guided resections based on clinicopathologic variables has not been explored. Our aim in this study was to prospectively evaluate whether the histopathologic expression of plasma membrane glycoprotein folate receptor analogs was predictive of the fluorescent localization of lung nodules during IMI-guided resections.

The increasing adaptation of various societal high-risk patient population screening guidelines has invariably led to a rising incidence of indeterminate lung nodules [24]. A substantial number of these patients will need some sort of tissue sampling for diagnosis, staging, and treatment. There are considerable institutional and interprovider-driven differences that complicate this matter. Standard of care in our institution does not require preoperative tissue sampling prior to index lung cancer resection by thoracic surgeons, but are subject to change with the recent approval of pafolacianine by the FDA. Therefore, in our previous large reports on pafolacianine-guided resections, IHC analyses were performed primarily 1–2 weeks after intraoperative findings. This has allowed us to identify correlations with FRα and FRβ, but until now, we have not analyzed whether knowledge of the plasma membrane protein density can predict intraoperative fluorescence. Thus, we wanted to select patients who had preoperative tissue sampling and analyze its predictive ability.

Our small cohort of patients was representative of the general surgical population treated with IMI. These patients tended to be Caucasian females with a significant smoking history with lesions noted to be FDG positive on PET imaging. Interestingly, we did notice more right-side predominant lesions in this cohort, but there was no change in tumor biology or tracer distribution. All patients received the same dose of pafolacianine within 24 h of surgery and did not have alterations to the standard of care curative surgery despite the prospective validation (Table 1). More than ¾ of the patients who were diagnosed with adenocarcinoma also served our preliminary analysis well. Adenocarcinomas have strong expression of FRα (Figs. 3, 4, and 5) on their surface, which would help us validate our findings and reach statistical significance, whereas the presence of other lesions, including benign findings, served as an internal negative control. It should be noted that inclusion of benign nodules in the study was because the radiological parameters were concerning despite the diagnosis and/or patients had more than lesions noted on imaging, but only select ones were amenable for preoperative sampling. Overall, the cohort of patients included in our study matched our previous experience and can potentially be generalized to our indeterminate lung nodule population with peripheral or peribronchial lesions amenable for sampling.

The advancement of screening high-quality cross-sectional imaging has improved the detection and characterization of nodules. As such, collaboration of chest imaging experts with thoracic oncologists has led to an increase in malignancy resection and a decrease in unnecessary parenchymal resections in patients with marginal pulmonary reserves [24, 25]. Despite the institutional practice pattern of not performing tissue sampling preoperatively, there was a > 89.1% diagnosis of malignant nodules in our surgical population. However, high-quality radiologic localization does not translate to intraoperative oncologic success, particularly for ground glass opacities. These lesions are not palpable and visually occult. In the past, the surgeon had to remove the entire lobe to ensure removal of the concerned lesion. IMI has allowed the detection of these lesions, which often convert full anatomic resection to nonanatomic sampling, leading to preserved lung function as well as decreased complications associated with lobectomies [4].

In our cohort, 29/30 malignant lesions were localized with pafolacianine-guided IMI resections. This included squamous cell carcinomas as well. However, the greatest fluorescence emission intensity was most prevalent in adenocarcinoma patients, with a median TBR of 3.11 ± 0.31, whereas SCC lesions had a TBR of 1.89 ± 0.29. Benign lesions did not demonstrate fluorescence and had similar fluorescence intensity as normal lung parenchyma (Fig. 1A-D). Analysis of these lesions demonstrated cellular level–specific labeling by pafolacianine with only tumor cell uptake of the tracer on fluorescence microscopy (Fig. 1). Interestingly, while FDG PET SUV levels have been found to correlate with the metabolic activity of tumors and are associated with malignancy risk, we did not observe the presence or lack of pafolacianine fluorescence to be associated with PET SUV levels, suggesting that potentially metabolically quiescent cells will uptake folate given the receptor density without glucose uptake. This observation is beyond the scope of our aims and needs to be further studied in a prospective manner.

Given that pafolacianine is a folate-conjugated fluorochrome and its fluorescence localization is correlated with folate receptor expression, we explored whether preoperative tissue expression of folate receptor analogs (alpha and beta) predicted intraoperative fluorescence detection and emission intensity. FRα is expressed primarily in tumors, including lung adenocarcinoma, ovarian cancer, and breast adenocarcinoma, and is the primary target of pafolacianine, whereas FRβ is found in tumor-associated macrophages, which are recruited to the periphery of tumors as part of an inflammatory response. Labeling either or both analogs pertains to an advantage to pafolacianine. For example, if the tumor is highly necrotic and FRα sampling is negative due to non-viable cells, the presence of a strong immunomodulatory response around the tumor can still label the tumor on NIR imaging (Supplementary Fig. 1).

Similar to our previous retrospective observations, the presence of FRα was strongly associated with nodule fluorescence, with the highest TBR observed in lesions with the highest density FRα on immunohistochemical analysis [6, 26]. Lack of any FRα on IHC corresponded with lack of fluorescence and TBR of 1.23 ± 0.22 (p < 0.05) (Fig. 2). FRα presence with TBR > 2 had an R² of 0.71, which is moderate to strong correlation. Squamous cell carcinomas that did not fluorescence intraoperatively similarly had low to minimal expression of FRα on preoperative IHC versus SCCs that fluoresced, which had FRα IHC scores of 1 + and 2 + . This near linear and strong association was not evident when analyzing samples for FRβ. There has been strong enthusiasm for FRβ cross-reactivity and pafolacianine labeling of FRβ + lesions. This is because pafolacianine has been observed to fluoresce in benign granulomatous lesions with strong FRβ density as well as benign lymph nodes, which demonstrate false positivity intraoperatively secondary to 3 + FRβ density (Supplementary Fig. 1, Fig. 4). Only in cases with strong 3 + IHC scoring for FRβ was there a correlation with fluorescence detection. It is likely that patients with lesions were malignant with a lack of FRα staining but strong 3 + FRβ staining positivity could benefit from pafolacianine-guided resections. However, given our observations in this small cohort, FRα has a stronger correlation and positive predictive value than FRβ staining levels. Larger patient populations can discern clearer correlations in the future.

Given that preoperative folate receptor staining can inform intraoperative fluorescence, we wanted to explore whether one sample biopsy was enough to make these observations. Although we had a small cohort in this exploratory group of patients, including low number of patients with benign diseases, we observed several key features. In large part, the majority of samples had similar IHC scores with excellent interobserver agreement (kappa 0.91) among the 9 independent reviewers. Each patient had 5 different passes and 5 different core biopsies performed for each lesion, and in patients with biopsy-proven malignancy, 4 had complete agreement among the samples. However, one sample from patient #5 (Fig. 5) had an absence of FRα, whereas the previous four had strong 3 + scoring. If this sample was the only taken, then this patient would be wrongfully deemed not a successful candidate for pafolacianine when in fact other samples show otherwise. Second, in the era where neoadjuvant immunotherapeutic decisions are made upon IHC staining, false negative results can be disease and life altering [27]. While the numbers are small to justify a recommendation, this phenomenon needs to be further investigated to determine if more than biopsy is necessary for preoperative stratification of pafolacianine infusion, especially with the recent FDA approval pafolacianine for ovarian and lung cancer resections.

There are several limitations in this study that need to be highlighted. While this is one of the first prospective validatory studies evaluating the role of preoperative folate receptor staining for patient selection in pafolacianine-guided lung nodule resections, the overall number of patients is small and does not include a substantive number of non-adenocarcinoma patient cohort. A larger patient population should be explored to minimize the effect of confounders and observe correlations, especially in settings where there are dichomatous variables such as FRα and FRβ density levels. Furthermore, FRβ expression is confounded by various factors including tumor necrosis secondary to tumor-associated macrophages and anthracotic lesions which should be accounted for when evaluating patients [28]. Additionally, we explored the role of one tracer for one type of cancer (lung cancer), and these preoperative findings beyond pulmonary malignancies are not known. Our findings, however, demonstrate for the first time that in a select group of patients, preoperative sampling can be predictive of intraoperative success. These claims should be further confirmed in a larger clinical setting in a randomized manner of all surgical eligible lung nodule patients to further discern the predictive ability of FRα for pafolacianine NIR labeling of lung nodules intraoperatively during IMI-guided resections, especially in the setting of recent approval of pafolacianine for use in lung cancer management by the FDA [7].

Conclusion

The results of our small cohort prospective preoperative analysis demonstrate that immunohistochemical staining for FRα has a strong concordance with pafolacianine fluorescence during IMI-guided lung cancer resections. Only strong FRβ positivity has been associated with the detection of fluorescence and can be particularly useful in non-adenocarcinoma lesions. Knowledge of these observations can inform the surgeon in selecting the optimal patient who would benefit from pafolacianine-guided resections. Further large-scale prospective studies are needed to validate these observations, as this proof of principle study evaluated predominantly adenocarcinoma spectrum lesions with minimal inclusion of SCC and benign lesions.

Data availability

All data is available in the manuscript and the supplementary file. Additional data can be obtained by contacting the corresponding author.