Pafolacianine for intraoperative molecular imaging of cancer in the lung: The ELUCIDATE trial

Abstract

Objective:

The study objective was to determine the clinical utility of pafolacianine, a folate receptor–targeted fluorescent agent, in revealing by intraoperative molecular imaging folate receptor α positive cancers in the lung and narrow surgical margins that may otherwise be undetected with conventional visualization.

Methods:

In this Phase 3, 12-center trial, 112 patients with suspected or biopsy-confirmed cancer in the lung scheduled for sublobar pulmonary resection were administered intravenous pafolacianine within 24 hours before surgery. Participants were randomly assigned to surgery with or without intraoperative molecular imaging (10:1 ratio). The primary end point was the proportion of participants with a clinically significant event, reflecting a meaningful change in the surgical operation.

Results:

No drug-related serious adverse events occurred. One or more clinically significant event occurred in 53% of evaluated participants compared with a prespecified limit of 10% (P <.0001). In 38 participants, at least 1 event was a margin 10 mm or less from the resected primary nodule (38%, 95% confidence interval, 28.5–48.3), 32 being confirmed by histopathology. In 19 subjects (19%, 95% confidence interval, 11.8–28.1), intraoperative molecular imaging located the primary nodule that the surgeon could not locate with white light and palpation. Intraoperative molecular imaging revealed 10 occult synchronous malignant lesions in 8 subjects (8%, 95% confidence interval, 3.5–15.2) undetected using white light. Most (73%) intraoperative molecular imaging–discovered synchronous malignant lesions were outside the planned resection field. A change in the overall scope of surgical procedure occurred for 29 of the subjects (22 increase, 7 decrease).

Conclusions:

Intraoperative molecular imaging with pafolacianine improves surgical outcomes by identifying occult tumors and close surgical margins. (J Thorac Cardiovasc Surg 2023;166:e468–78)

Graphical Abstract

Lung adenocarcinoma imaged intraoperatively with pafolacianine and near-infrared light.

Surgical resection is the best option for patients with stage I non–small cell lung cancer (NSCLC); however, there is still a 15% to 20% local failure rate within the first 5 years.^1–3 Locoregional recurrences occur because surgeons are unable to completely detect and remove the primary tumor nodule, synchronous lesions are not visualized on preoperative imaging or at the time of surgery,^4–6 or disease is left behind at resection margins.^7,8 An increased margin distance 15 mm or more has been shown to be associated with a lower risk of local recurrence of NSCLC, with a 10-mm margin distance having a 45% lower recurrence risk than a 5-mm distance.⁹ Resection of the target lesion with sufficient margin and identification of synchronous cancerous lesions are vitally important to avoid cancer recurrence. These challenges have become magnified because of the increasing use of minimally invasive surgery (thoracoscopic and robotic surgery) where tactile feedback is reduced and complete assessment potentially compromised. Another challenge is locating small lesions such as ground glass opacities, which are early-stage lung adenocarcinomas. Current transthoracic or endobronchial localization techniques, including computed tomography (CT)-guided and endobronchial interventions, require direct physical or imaging marker placement often with use of ionizing radiation, and can add significant procedure time and morbidity by procedural related hemorrhage, pneumothorax, sedation-related complications, and pain.¹⁰ Intraoperative molecular imaging (IMI)¹¹ has the potential to address these challenges.¹² IMI uses a targeted agent that optically fluoresces during surgery to draw attention to cancer cells. Tumor-specific accumulation allows identification with no a priori knowledge of the presence or location of the cancer. IMI permits immediately actionable postresection analysis of margins in the operating suite.

Pafolacianine, a folate analogue indocyanine green-like conjugate, is a novel fluorescent imaging agent that binds FR with approximately 1 nM affinity and eliminates from receptor-negative tissues with a half-life of less than 30 minutes.^13–15 By accumulating preferentially in FR + tumors, pafolacianine can label cancerous nodules so they are visually highlighted intraoperatively when excited using a near-infrared (NIR) lighting system (Figure 1), with minimal photobleaching.¹⁶ As much as 85% of pulmonary malignancies express folate receptors (FRs),^17–19 which are unaffected by chemotherapy.²⁰ In this prospective Phase 3 clinical trial, we hypothesized that IMI with pafolacianine improves surgical outcomes by reducing positive margins, identifying additional cancers, and localizing difficult to find lesions.

FIGURE 1.

Representative example of molecular imaging. A, Primary nodule of adenocarcinoma in the upper left lobe, measuring 162 mm³ in volume at a depth of 8.9 mm, visualized using white light only. B, The same nodule viewed under NIR.

MATERIALS AND METHODS

Trial Design and Oversight

Before surgery, participants were randomized to undergo white light plus IMI or white light only (10:1 ratio, as requested by the Food and Drug Administration). The assignment for imaging was revealed during surgery but only after completion of the white light evaluation and before IMI. The purpose of the 10:1 nonstratified permuted block randomization was only to reduce the risk of potential bias from underestimating the number of lesions localized first with white light if knowing that NIR would be used. Infusion of study drug was also administered before surgery to all participants. Participants exposed to the study drug and assigned to white light only were evaluated for safety only. The trial protocol and data publication were approved by an Institutional Review Board (November 25, 2019, # Pro00040361). Study oversight was provided by the funding company and a contract research organization. Written informed consent including data publication was obtained from each patient before initiating screening procedures. Enrollment was from June 2020 to September 2021. Twelve sites participated, including academic medical centers and regional hospitals. A variety of training methods including peer-to-peer observation, virtual proctorship, and training conducted by the camera manufacturer or sponsor, were completed for each of the investigators.

Participants

Consenting adult patients aged 18 years or more qualified for the trial if they had a diagnosis or high suspicion of cancer in the lung based on recent positron emission tomography/CT imaging (primary or metastatic) and were scheduled to undergo surgical thoracoscopy for a sublobar resection regardless of whether the primary nodule was expected to be visible or palpable in situ. FR expression testing of biopsy specimens was not required. Exclusion criteria included pregnancy, impaired renal function (estimated glomerular filtration < 50 mL/min/1.73 m²), impaired liver function (values > 3× the upper limit of normal for alanine aminotransferase, aspartate aminotransferase, or alkaline phosphatase, total bilirubin > 2× upper limit of normal, excluding participants with Gilbert’s syndrome), or clinically significant abnormalities on electrocardiogram at screening.

Trial Procedures (Figure 2)

FIGURE 2.

Methods, results, and implications are presented. IMI, Intraoperative molecular imaging; WL, white light; FS, full set; CSE, clinically significant event; NIR, near infrared.

Each participant was scheduled to receive 0.025 mg/kg pafolacianine intravenously over 60 minutes within 1 to 24 hours before surgery. Premedication for anticipated nausea or vomiting was allowed before infusion, based on prior studies. Each surgery began using white light visualization to attempt to identify and localize the target nodule(s). Once the white light visualization was completed, the sealed envelope assigned to the participant was opened to reveal whether surgery would proceed with or without additional exploration with NIR imaging using the Visionsense VS3 Iridium camera system (Medtronic). With the use of IMI, the lung was reinspected searching for fluorescence emanating from the primary nodule or other synchronous lesions (Video 1). Pulmonary resection was then performed under white light. The surgeon determined whether observed fluorescence was likely to indicate tumor based on visual assessment of the fluorescence signal strength, near-infrared appearance, and clinical judgment, and thus could elect to not resect what appeared to be false-positive fluorescence. The resected specimens in the IMI group were subsequently examined on the back table for measurement of surgical margins, which in some cases led to additional surgery. The surgeon illuminated the resected tissue with the NIR light system and, using a ruler, measured the distance between the nearest edge of the lesion to the closest staple line. Before completion of the procedure, the surgical field was inspected again with NIR to detect any additional fluorescent nodules, which were removed at the discretion of the surgeon. Participants were evaluated at days 7 and 28 after surgery for follow-up.

VIDEO 1.

Download video file ^{(12.4MB, mp4)}

Intraoperative video showing a lung adenocarcinoma synchronous lesion in the right upper lobe as visualized using white light alone (top left), NIR (top right), white light with false color NIR overlay, variant number 1 (bottom left), and white light with false color NIR overlay, variant number 2 (bottom right). Pafolacianine had been infused intravenously before surgery. Video available at: https://www.jtcvs.org/article/S0022-5223(23)00185-X/fulltext.

Outcome Assessments

Excised nodules and synchronous lesions were sent to the local pathology laboratory for histological examination, and, when available, tissue was also sent to a central laboratory (Moffitt Cancer Center) for immunohistochemical examination of FR expression. The full set (FS) for analysis of safety included all participants exposed to pafolacianine (n = 112). Spontaneous and elicited adverse events were collected for analysis of safety. Analysis of the primary efficacy end point included participants exposed to pafolacianine, who underwent surgery under both white light and NIR fluorescent imaging (IMI, n = 100).

End Points

The primary efficacy end point was the proportion of IMI participants with a clinically significant event (CSE), defined as at least 1 of the following.

• Removal of 1 or more primary “lung nodule(s)” detected only by NIR examination, not detected by white light or palpation, and confirmed by histologic examination to be other than benign lung parenchyma.

• Removal of 1 or more cancerous synchronous lesions detected only by NIR examination, not detected by white light or palpation, and confirmed by histologic examination to be cancerous.

• Identification of a cancerous positive margin 10 mm or less under NIR, assessed in the operating room by the investigator and confirmed by histologic examination.⁹

Secondary end points included calculation and estimation of sensitivity and false-positive rates (where False-Positive Rate = (False Positive)/(True-Positive + False-Positive)) for all 100 IMI participants with suspected or confirmed cancer in the lung, for those with a confirmed diagnosis of cancer, and for those with both FRα+ nodules/lesions. The sensitivity and false-positive rate for detecting FRα+ tissue was also determined. Part of the subset analysis included calculation of true-positive rates, meaning lesions found with IMI that were also cancer based on histopathology. These are a subset of the first bulleted CSE above. Time to localization and change in the scope of surgery were also evaluated. Safety end points were the incidence rates of all treatment-emergent adverse events, adverse device effects, and serious AEs, from the time of pafolacianine (OTL38) administration through follow-up visit 4; evaluation of laboratory parameters (chemistry and hematology) and vital signs; and evaluation of electrocardiograms before and after study drug administration.

Statistical Analysis

The prespecified analysis for the CSE end point was a 1-sample test comparing the observed proportion of IMI participants with 1 or more CSE against a prespecified threshold of 0.10. The analysis was conducted via an exact binomial test at the 2-tailed alpha level of 0.05 with accompanying exact (Clopper-Pearson) 2-sided 95% confidence intervals (CIs). The target sample size for the primary analysis was 100 IMI participants assuming a true CSE proportion 0.20 or more and 80% power. IMI participants whose CSE status could not be determined would be imputed with a non-CSE result. (No imputation was necessary.)

The prespecified statistical model for estimation of the sensitivity and false-positive rate was a generalized linear mixed model for a binomial outcome with a logit link function containing a random effect for participant and a fixed effect for the constant. The random effect was to allow for intra-cluster dependence among multiple tissue samples within the same participant. The model was to use maximum likelihood estimation with adaptive quadrature. However, the results produced very wide CIs, even when the model met the default convergence criteria for the software (SAS Version 9.4, SAS Institute Inc). This was believed to occur because only a few participants contributed more than 1 tissue sample. In fact, most participants contributed only a single specimen to the analyses. As a result, the same generalized linear mixed model using Proc GLIMMIX in SAS with maximum likelihood marginal expansions pseudolikelihood (ie, modified maximum pseudo-likelihood) estimation was run without convergence issues. These latter estimates are considered population-averaged or marginal estimates and are the model estimates presented. It was also prespecified that the crude, nonmodeled estimates also be calculated. These were observed to be very close to the modified maximum pseudo-likelihood estimates. Prespecified subgroup analyses for the primary and secondary efficacy end points examined self-reported sex, age, race, ethnicity, and smoking history, as well as study center, prior chemotherapy treatment, prior radiation therapy, prior cancer history, lung cancer histologic type, surgery types, infusion (day before vs day of surgery), time from stop of infusion to imaging start time, and prior lung cancer surgery. Safety end points were descriptively analyzed and included all participants exposed to study drug regardless of surgery status or group.

RESULTS

Characteristics of the Participants

The Consolidated Standards of Reporting Trials flow diagram is presented in Figure 3. Among 140 patients screened for eligibility, 24 were screen failures and 4 of the remaining 116 randomized participants were discontinued during the screening period before receiving an infusion of the study drug, and 1 additional participant withdrew consent because of an infusion reaction. Demographic characteristics of participants in the FS (N = 112), which includes the 1 participant who received the infusion but withdrew consent and the IMI (N = 100), are presented in Table 1. In the FS, 25 participants (22.3%) had a history of 31 cancerous lung tumors. Sites of origin of pulmonary metastases included colon and rectum, ovary, pancreas, and bone. The initial tumor stage was generally 1 or 2 disease. The median greatest length of the primary tumors imaged by IMI was 13 mm (range, 5–42 mm), whereas the median greatest length of the synchronous lesions identified by IMI was 12.2 mm (range, 2–52 mm).

FIGURE 3.

Consolidated Standards of Reporting Trials diagram. Flow diagram showing the enrollment of subjects, their allocation to treatment, their disposition status, and how they are analyzed in the trial. IMI, Intraoperative molecular imaging; WL, white light; FS, full set.

TABLE 1.

Demographic and clinical characteristics of the 112 participants

	IMI (N = 100)	WL (N = 11)	FS (N = 112)*
Sex, no. (%)  Male  Female	39 (39.0) 61 (61.0)	2 (18.2) 9 (81.8)	41 (36.6) 71 (63.4)
Childbearing potential, no. (%)	1 (1.6)	0	1 (1.4)
Age (y)  Mean  SD  Median  Min, Max	66.0 9.68 67.0 26, 83	64.4 9.88 66.0 44, 79	65.9 9.66 67.0 26, 83
Age (categorized), no. (%)  <65 y  ≥65 y	36 (36.0) 64 (64.0)	5 (45.5) 6 (54.5)	41 (36.6) 71 (63.4)
Ethnicity  Hispanic or Latino  Not Hispanic or Latino  Not reported	0 99 (99.0) 1 (1.0)	0 11 (100) 0	0 111 (99.1) 1 (0.9)
Race, no. (%)  Asian  Black or African American  White  Unknown  Other	2 (2.0) 8 (8.0) 88 (88.0) 1 (1.0) 1 (1.0)	0 3 (27.3) 8 (72.7) 0 0	2 (1.8) 11 (9.8) 97 (86.6) 1 (0.9) 1 (0.9)
Smoking history, no. (%)  Current  Former  Never	19 (19.0) 52 (52.0) 29 (29.0)	2 (18.2) 6 (54.5) 3 (27.3)	21 (18.8) 59 (52.7) 32 (28.6)

IMI, Imaging plus WL; WL, white light only; FS, all participants exposed to pafolacianine; SD, standard deviation.

^*One participant withdrew consent after the infusion.

Primary Outcome

For the primary CSE end point, 53 of 100 IMI participants (53%) achieved at least 1 CSE, which was significantly greater than the prespecified 10% threshold (95% CI, 42.8–63.1, P<.0001). Forty-three participants had only 1 CSE and the other 10 had more than 1 with a total of 65 CSEs occurring among these 53 participants. Nineteen of the CSEs comprised identification of the primary nodule in 19 of the 100 participants (19%) that could not be located by white light or palpation examination. NIR imaging identified a total of 10 synchronous cancers in 8 participants not detectable by preoperative imaging or intraoperative white light inspection: 7 adenocarcinomas, and in 1 participant 3 metastatic chordomas. In all these cases, the additional cancers were removed with simple wedge resection. Only 8 of these synchronous lesions were counted as CSEs with 3 synchronous lesions in 1 participant being considered as 1 CSE. Of note, 4 of these 8 participants had synchronous lesions in a different lobe from the primary; thus, they would have been missed by conventional surgery. The remaining 38 of 65 CSEs consisted of identification of fluorescence within 10 mm or less of the closest resection margin as measured by the surgeon using back-table NIR imaging. The investigator measurement obtained real-time in the operating room was required for purposes of evaluating the margin distance in the trial. Measurements taken by a blinded local pathologist were not standardized for purposes of the trial but followed local standard procedures. Post hoc analysis found that in 32 of these 38, local pathology measurement was in agreement with the surgeon’s measurement using NIR, consistent with prior experience.²¹ There were 40 additional tissues where back table measurements under NIR showed margins greater than 10 mm (not a CSE), 15 of which were measured by the pathologist as 10 mm or less. Among the 10 participants having more than 1 CSE, 1 had both an unidentified primary lesion (unidentified under white light examination with palpation) and a synchronous lesion; 5 had an unidentified primary lesion as well as a close margin; 2 had additional synchronous lesions identified and close margins; and 2 other participants had all 3 CSEs (Table 2).

TABLE 2.

Clinically significant events

Participant level CSEs overall	Participants	% of 100 evaluable participants
Participants with ≥1 CSEs	53	53.0% 95% CI (42.8–63.1) P <.0001 (vs 10% threshold)
Participant level CSE by type	Participants	% of 53 CSE participants
Primary only (P)	11	20.8%
Synchronous lesion only (SL)	3	5.7%
Margin only (M)	29	54.7%
P, SL, M	2	3.77%
P, SL	1	1.89%
P, M	5	9.43%
SL, M	2	3.77%
Individual CSEs by category	CSEs	% of 65 CSE
Primary lesion	19	29.2%
Synchronous lesion	8	12.3%
Margin	38	58.5%
Margins by distance	Margins	% of 38 positive margins
≤5 mm	23	61%
>5–10 mm	15	39%

CSE, Clinically significant event; CI, confidence interval.

Secondary Outcomes

A total of 134 specimens were excised from the 100 IMI participants, with each participant contributing 1 or more specimens. All were sent for local histopathology, with 104 specimens found to be positive for cancer in 89 participants. Tissue was available from a subset of 113 specimens that were sent to the central laboratory for FRα staining, which was positive in every case. Among these 113 FRα+ specimens, 92 were cancer. Among all 134 specimens from participants with suspected or confirmed cancer, 108 (81%) had fluoresced under IMI in 78 participants. The estimated sensitivity for detecting a cancerous tissue was 80/104 or 76.9% (model estimate 76.5%; 95% CI, 66.7–84.2). There were 28 of 108 (25.9%) false-positives (10 primary nodules, 18 synchronous lesions). Histology on the false-positive tissues was mostly benign or normal lung parenchyma. Where pathology was identified, it was most often granulomatous disease, with 1 fibrous tumor, 1 meningothelial-like nodule, 1 anthracotic nodule, and 1 lipoid pneumonia. The model estimate for the false-positive rate was 25.8% (95% CI, 18.2–35.2). (Table 3). In participants with suspected or confirmed cancer and known positive FR expression, the sensitivity was 71/92 or 77.2% model estimate 76.8% (95% CI, 66.3–84.8) with a false-positive rate of 19/90 or 21.1% (model estimate 21.1% (95% CI, 13.8–30.9). Among the 89 participants with confirmed cancer regardless of FR expression, 80 of 104 resected cancerous lesions were detected using NIR imaging resulting in a sensitivity of 76.9% (model estimate 76.5%; 95% CI, 66.7–84.2). More specifically, NIR imaging identified 97 fluorescent lesions in 69 of these 89 participants, with 17 of 97 (17.5%) lesions proving to be false-positive. The model estimate of false-positivity rate among these participants with confirmed cancer was 17.5 (95% CI, 11.1–26.6) with a rate for primary nodules of 1.4% (95% CI, 0.2–9.8) and for synchronous lesions of 59.7% (95% CI, 37.5–78.4). The sensitivity for detecting cancerous primary nodules and synchronous lesions in participants with confirmed cancer and known positive FR expression was 71 of 92 or 77.2% (model estimate 76.8%; 95% CI, 66.3–84.8) and the false-positive rate was 11.3% (model estimate also 11.3%; 95% CI, 5.9–20.5). Among the 113 FRα+ specimens, 91 had fluoresced under IMI. The sensitivity for detection of FRα+ lesions in participants with suspected or confirmed cancer in the lung was 91/113 or 80.5% (model estimate 80.0%; 95% CI, 70.8–86.8), and none of the specimens were false-positive. The median time to localize the primary nodule was 1 minute (range <1–23) using IMI after white light examination, compared with a median time 4 minutes (range, <1–48) during white light examination, including cases where no lesion was found under white light only. The median depth of the primary lesions detected by IMI only was 10.1 mm (0–27.9 mm) as measured by preoperative CT/positron emission tomography imaging, and for lesions detected by both white light and IMI 2.3 mm (0–37.7 mm). Among primary lesions seen with IMI at depths of 10 mm or greater (n = 9), the average volume was 6000 mm³ (range, 75–27,436) versus 2293 (range, 45–18,900) for more superficial lesions. Among 11 participants whose primary nodule did not fluoresce in situ, 45.5% (5/11) had the primary nodule identified by back-table NIR. Cases in which the lesion was found only using white light and palpation were all wedge resections and identified as suspicious for cancer on presurgical CT or positron emission tomography scans. Ten cancerous synchronous lesions were found in 8 participants only using IMI, with 8 of the 10 (80.0%) in an anatomic portion of the lung not planned for surgical resection. Metastatic disease from the following sites of origin were imaged: colorectal, breast, renal, melanoma, spine, ovarian, prostate, pancreatic, liposarcoma, and neuroendocrine. Lesion visualization with NIR overall resulted in an increase in the scope of the surgical procedure for 22 participants and a decrease for 7. The results of subgroup analyses, including operative procedure, underlying pathology, and clinical stage, were not informative or different from the full set analyses.

TABLE 3.

Estimated sensitivity and false-positive rates for cancerous tissue

	Participants with suspected or confirmed cancer in the lung (n = 100)		Participants with confirmed cancer in the lung (n = 89)
	Sensitivity	False-positive	Sensitivity	False-positive
Total No. of participants with evaluated tissues	89	78	89	69
Total No. of evaluated tissues	104	108	104	97
Overall accounting for intra-cluster	76.5% (66.7–84.2)	25.8% (18.2–35.2)	76.5% (66.7–84.2)	17.5% (11.1–26.6)
(estimated % and 95% CI)
Primary nodules	74.2 (63.7–82.4)	12.7 (6.9–22.3)	74.2 (63.7–82.4)	1.4 (0.2–9.8)
Synchronous lesions	-*	62.6 (41.2–80.0)	-	59.7 (37.5–78.4)
Cancer histological type accounting
for intra-cluster
Adenocarcinoma	77.3 (65.6–85.8)	-	77.3 (65.6–85.8)	-
Squamous cell carcinoma	83.0 (17.6–99.1)	-	83.0 (17.6–99.1)	-
All other cancers	69.4 (42.7–87.4)	66.7 (48.7–80.9)	69.4 (42.7–87.4)	54.3 (34.4–72.9)
Nonmalignant	-	28	-	17

^*Model estimates are missing when there is no variability. For synchronous lesion sensitivity, 10 of 10 cancerous lesions were detected. For adenocarcinoma/squamous cell carcinoma false-positive rates, 0 of 58 and 0 of 5 detected lesions were false-positives.

Extent of Exposure

Most (106/112; 94.6%) participants received the full dose, 3 (2.7%) received a partial dose, and for 3 (2.7%) dosing information was missing. The infusion was interrupted for 12 participants (10.7%). For those who had infusion interruptions, the infusion was restarted once for 6 participants (50%) and twice for 1 participant (8.3%). Of the 12 participants who had infusion interruption, 7 (58.3%) ultimately received the complete dose. The mean (standard deviation) total infusion duration was 60.6 (13.29) minutes with a median of 60.0 minutes for participants overall. The time from end of infusion to NIR illumination ranged from 1 to less than 6 hours for 46 participants (41.1%), from 6 to less than 12 hours for 5 participants (4.5%), and from 12 to 24 hours for 45 participants (40.2%). The median total exposure time to fluorescent light was 2.0 (0–23) minutes. There were no statistically significant differences in CSE based on dosing time.

Safety

No drug-related serious adverse events occurred. Apparent reactions were observed in 19 participants (17.0%), including 4 of 42 (9.5%) who had been premedicated. The most frequently reported infusion reactions included nausea (10 participants), intermittent hypertension (4 participants), vomiting (3 participants), and abdominal pain, flushing, and transient hypotension (2 participants each). Pruritus and urticaria were each seen in 1 participant. These reactions led to infusion interruption in 11 participants and infusion termination in 5 participants. Most of the reactions were considered mild or moderate in intensity and resolved by the end of the infusion, and all resolved by the end of the study. The proportion having 1 or more CSE (9/16, 56%) was not diminished. A total of 42 participants received premedication with diphenhydramine, typically together with ondansetron, granisetron, or famotidine, for prophylaxis of possible infusion reactions.

Overall, 524 adverse events were recorded for 110 of 112 participants who were exposed to pafolacianine. The vast majority were unrelated to IMI with pafolacianine and related to the surgical procedure itself. The most commonly reported were injury/procedural complications (91.1% of participants, including expected postoperative procedural pain [79.5%]), respiratory, thoracic, and mediastinal disorders (54.5%), including pneumothorax (26.8%), cough (17%), dyspnea (17%), hypoxia (11.6%), pleural effusion (8.0%), and pulmonary air leakage (5.4%), gastrointestinal disorders (33.9%, including nausea in 19.6%), and vascular disorders (19.6%). Participants who had false-positive synchronous lesions removed showed no associated increase in respiratory or pulmonary adverse events.

DISCUSSION

Pafolacianine represents a first-in-class agent to aid the surgeon in visualizing otherwise undetectable tumors and in determining tumor to resection margin distance.^20,22 It offers the possibility of improving outcomes of cancer surgery by enabling surgeons to better identify primary tumors, detect occult synchronous lesions, and ensure adequate margins of resection. The technology is adjunctive to, rather than a replacement for, current preoperative imaging and intraoperative clinical judgment. Pafolacianine binds quickly to the targeted receptor and clears sufficiently rapidly from nonmalignant tissues to allow tumor visualization with an optimal tumor-to-background contrast ratio.²³ The combination of high affinity receptor binding and internalization via receptor-mediated endocytosis, selective concentration to FR + cancer tissues, and a short clearance time from receptor-negative compartments allows infusion the day before or the day of surgery.²⁴ Compared with other existing fluorescent imaging agents, pafolacianine shows less auto-fluorescence due to its near-IR excitation wavelength and benefits from a meaningful NIR depth of penetration.^23,25 In this trial, time to identification of the primary nodule was shortened, and a high sensitivity for lesion detection was associated with a low false-positive rate. Although the most frequent CSE was identification of a close margin, identification of an otherwise undetectable primary nodule occurred in 19% of participants. None of the previously unnoticed, fluorescent synchronous lesions were identified by a central reader on retrospective review of CT scans. The importance of IMI in central tumors is for the possibility of synchronous tumors in other lobes. With regard to spread through airway spaces disease, these tend to be adenocarcinomas; thus, pafolacianine would be effective. By consensus among the investigators and Food and Drug Administration, it was agreed a priori that finding a CSE in 10% or more of instances would be sufficiently meaningful. Among clinicians, it may be that any improvement toward removing cancerous tissue is of value.

Pafolacianine was generally well tolerated, with no major safety concerns identified and no serious adverse events related to study drug. A small number of manageable infusion reactions occurred, the mechanism of which is unknown at this time.

Study Limitations

All sites used the same imaging system. In patients with stage II or III NSCLC, nodal recurrence or in-transit lymphatogenous metastases are also factors in local recurrence. These patients will typically receive postoperative chemotherapy or radiotherapy; thus, the recurrence rate is difficult to tease out from surgery versus the other therapies. IMI is likely to be of greatest benefit in cases where sublobar resection is used for early-stage lung cancers to ensure adequacy of margins and identification of second primary tumors. Conversely, early intraoperative identification of occult lung or pleural metastases in patients with more advanced tumors conceivably could be of benefit in avoiding potentially futile surgery. Potential improvements in care may come not only from more complete surgical resection but also from upstaging of the cancer in some cases that would lead to appropriate adjuvant therapy. Although IMI showed improvements over white light assessment, it has not yet been evaluated against alternative localization modalities such as percutaneous or endobronchial localization. The value of pafolacianine in mediastinal lymph node dissection was not examined in this study. Further investigation is needed to extend the findings to anticipated improvements in disease-free survival.

CONCLUSIONS

The use of IMI with pafolacianine represents a significant potential advancement over white light visualization alone by enhancing the intraoperative localization of presumed cancerous nodules, improving the ability to remove them with clean margins, and reducing the probability of leaving malignant tissue behind.

PERSPECTIVE.

Five-year survival remains low after lung cancer surgery. Recurrences may arise from missed primary nodules, cancer synchronous lesions not seen on preoperative imaging or intraoperative examination, and inadequate surgical margins. Intraoperative real-time imaging techniques may increase the accurate identification of cancerous lesions, limit resections, and potentially improve survival.

CENTRAL MESSAGE.

IMI with pafolacianine represents a significant potential advancement over current standards of surgery for cancer in the lung.

Abbreviations and Acronyms

CI

confidence interval

CSE

clinically significant event

CT

computed tomography

FR

folate receptor

FS

full set

IMI

intraoperative molecular imaging

NIR

near-infrared

NSCLC

non–small cell lung cancer

Biographies

Presenter: Dr Sunil Singhal

Dr Kazuhiro Yasufuku (Toronto, Canada). I congratulate Dr Singhal and the co-investigators on completion of a phase III, open-label, 12-center trial in the midst of COVID of IMI of FR-positive cancer in the lung, using pafolacianine. I have enjoyed seeing how Dr Singhal has been able to translate this first-in-class agent to aid thoracic surgeons in visualizing these tumors during sublobar resections. I think you are taking this to the next level, and it is almost ready for clinical practice. What was the rationale for the 10:1 randomization? There seems to be significant imbalance between the white light and IMI groups.

Dr Sunil Singhal (Philadelphia, Pa).I agree, the randomization was peculiar at 10 to 1, IMI versus no treatment. I would like to make 2 points. First, each patient acted as their own control. The surgeon entered with the thoracoscope and then did the best operation possible. The surgeon would look around with white light first and recorded any findings. Only then did the surgeon turn on the camera for those patients randomized to near IMI. Therefore, each patient was their own control. The randomization plan was decided by the Food and Drug Administration. The investigators did not know the randomization scheme. The reason the Food and Drug Administration wanted randomization at all is because there was concern that if every patient received imaging, then the surgeons would not do the best inspection possible. Because the surgeon did not know if the patient was going to get imaging, this forced the investigators to do the best possible attempt to look around with the white light.

Dr Yasufuku. For your imaging system, I believe you used a commercially available system from Medtronic. Was that altered in any way so that you can actually see or were you using just a regular device that was available?

Dr Singhal. This is a special camera that has to be specifically designed to detect the dye called S0456. It fluoresces at 775 nm. This agent cannot be recognized by a NIR endoscope that detects indocyanine green. It has to be specifically designed just to visualize this dye.

Dr Yasufuku. That is great to know. For my last question, despite all tumors being reported as FR positive—I think your IMI in about 20% of identified nodules were also false-positive. Is there any suggestion on how to better select patients for IMI to maximize its utility?

Dr Singhal. The false-positives and false-negatives are going to continue to be a problem with this agent. First, there will continue to be false-negatives because the lesions are too deep to be visualized. That is the Achilles’ heel of IMI. The flip side is you don’t have to worry about radiation or safety in the operating room because this is a low energy technology. With regards to the false-positives, FR is present on macrophages. This causes granulomas, inflammatory macrophages, and necrotic squamous cell carcinomas to fluoresce. Thus, making decisions with imaging is not done in a vacuum. The data from IMI have to be combined with the decision-making by the surgeon. Part of it is thinking, “That looks like a granuloma. I don’t think I’m going to take that.” At this point, I believe this technology is an adjunct to everything else we do in the operating room to make decisions. It will not replace other technologies; rather, this is an adjunct to what we already have.

Dr Joseph B. Shrager (Stanford, Calif). Sunil, congratulations on your dedication to this over your whole career in bringing this technology forward; we’ve been talking about it from time to time at meetings and in person for years. You’ve heard some of these questions, but for the group, first, marking was not allowed. So you’re not really comparing it with what many people are using as the current technology. Correct?

Dr Singhal. Correct. This is tough study to do. There were 12 centers. Some centers use bronchoscopic indocyanine green installation versus needle localization versus fluoroscopy versus image-guided video-assisted thoracoscopic surgery. It is hard to standardize the other arm for this study.

Dr Shrager. How many of the second nodules identified were things that in retrospect were important to resect—the additional things found, that is. Or how many were in situ or minimally invasive adenocarcinoma that normally we would follow and maybe they would turn into something eventually or maybe they would not?

Dr Singhal. Eight of the additional tumors discovered by IMI were Stage 1A adenocarcinomas. We did not count in situ adenocarcinomas. What really surprised me was that a number of the additional cancers were poorly differentiated adenocarcinomas on final pathology. Many of us, for example, when performing a right upper lobe for a 2-cm right upper lobe adenocarcinoma are aware of a little wispy thing in the right lower lobe. We may say to ourselves, “Ah, well, it’s got a 5-mm solid component. It’s been there for 2 years. I am going to leave it alone.” In fact, these nodules are dangerous, and as long as you are there, you should consider removing them so long as you are not dramatically changing the magnitude of the operation.

Dr Shrager. That’s really surprising.

Dr Singhal. Yes, very interesting finding.

Dr Shrager. All of these patients got fine-cut CTs before-hand? Because it’s amazing to me that you can’t see all these things before on a fine-cut CT, 1 mm.

Dr Singhal. Everybody got a 1 mm fine-cut CT within 6 weeks. When you go back, you can sometimes retrospectively locate them.

Dr Shrager. Right. So if somebody’s really in-tune to looking carefully at where they are, I find it hard to believe that any ground-glass opacities can be missed by a fine-cut CT.

Dr Singhal. Yes, but so—

Dr Shrager. These were high-end investigators here who should certainly know how to see small ground-glass opacities on CT scans preoperatively, so I’m surprised.

Dr Singhal. I want everyone to remember the essence of this technology. It is to draw the attention of the surgeon to a lesion that may be overlooked or a margin that may be overlooked. It might draw your attention to a false-positive such as a granuloma, but it might also draw your attention to another cancer. Or it might draw your attention to a positive staple line margin after a wedge or segment. IMI is an adjunct to focus the surgeon’s eyes on something else that you may have ignored.

Moderator. It’s important to know: How close were the nodules to the periphery? In other words, were you saying you couldn’t get 10 mm away? What did you choose?

Dr Singhal. We could identify tumors in the outer one-third of the lung, typically at 2-cm depth.