Second Window Indocyanine Green: Near-Infrared Optical Contrast for Intraoperative Identification of Brain Tumors

12 Second Window Indocyanine Green: Near-Infrared Optical Contrast for Intraoperative Identification of Brain Tumors

Ryan D. Zeh, Ryan D. Salinas, Sunil Singhal, and John Y.K. Lee

Abstract
While indocyanine green (ICG), an FDA-approved near-infrared contrast agent, has classically been used in neurosurgery for visualization of vasculature, new techniques can permit surgeons to use ICG to detect tumors in vivo.1 Coined “second window” ICG (SWIG) in order to discriminate it from traditional videoangiography procedures that visualize the molecule within minutes of injection, this method relies on visualization 24 hours following high-dose intravenous infusion. A growing body of work has utilized this technique to visualize both primary and metastatic brain tumors. These preliminary findings suggest that SWIG has the potential for broad applications within the neurosurgical field as it can not only aid in localization intraoperatively, but also has promising features that can assist in margin detection.

Keywords: fluorescence imaging, indocyanine green, near-infrared, imaging, second window ICG

12.1 Introduction to Second Window Indocyanine Green

Indocyanine green (ICG, C₄₃H₄₇N₂NaO₆S₂) is a tricarbocyanine dye with a molecular mass of 751.4 Da that fluoresces in the near-infrared (NIR) range, with peak emission and excitation at 780 and 810 nm, respectively. ICG has been used in the medical field since the World War II era, and it was granted Food and Drug Administration (FDA) approval in 1959 for ophthalmologic applications.^1,2,3

The most common purpose for ICG has traditionally been for angiographic procedures with some application for vascular studies in brain tumors.⁴ Doses between 0.2 and 0.5 mg/kg are delivered to the patient intraoperatively; NIR light is then used to excite the dye, and an NIR camera is used to capture the emission of the molecule flowing through the vasculature for minutes following delivery of the bolus.

Early preclinical work with ICG has demonstrated that injection of ICG into rats can lead to persistence of signal an hour after injection. In 1996, Haglund et al found that ICG closely correlated with tumor margins in malignant rat tumors.⁵ Preliminary studies in humans also demonstrated similar results. These studies, performed in the 1990s, were not studying the ICG signal in vivo. In these works, the tumor margins were analyzed after resection. This limited the researchers’ ability to guide resection.

However, the use of ICG in human trials has only recently developed with the advent of novel endoscopes and open-field exoscopes that are able to facilitate intraoperative identification of ICG positive tissues following bolus injections of the molecule. A growing body of recent research has utilized ICG in a different manner from investigations of the vasculature. New techniques have been developed that can help surgeons localize tumors intraoperatively. This technique has been coined second window ICG (SWIG).⁶ In this procedure, doses as high as 5.0 mg/kg are delivered to the patient up to 24 hours in advance of imaging.

By delivering these high doses with injections nearly 24 hours prior to imaging, it is presumed that the ICG becomes subject to the enhanced permeability and retention (EPR) effect, the mechanism by which very small molecules (e.g., nanoparticles) can accumulate in tumors, or other places with similarly unique pathophysiological characteristics.⁷ While the exact mechanism is not entirely clear, it is believed that ICG reversibly binds to serum albumin. Potential contributing factors to its subsequent deposition in tumors include defective endothelial cells, fenestrations in the vasculature, disorganized draining (e.g., lymphatic system), and otherwise altered permeability.^6,7,8 SWIG has demonstrated the capacity to identify solid tumors in subcutaneous animal models, canine, and in human clinical trials. Furthermore, studies of ICG in human tumors have been applied in numerous different cancer types not limited to lung, prostate, breast, ovarian, colorectal, pancreatic, esophageal, metastatic, and brain cancers.

12.2 Preclinical Work

Jiang et al demonstrated the value of delayed, high-dose SWIG in a rodent model, thus providing the basis for administering ICG to human patients 24 hours in advance of their surgery. In this study, ICG was administered to mice with subcutaneous tumors grafted from metastatic lung cancer cell lines, mesothelioma cell lines, and esophageal carcinoma cell lines. Varying doses of ICG ranging from 0.7 to 10.0 mg/kg were administered via tail vein injection to the mice. The tumors were then imaged serially at time points up to 72 hours later. In doing this, it was determined that 5.0 mg/kg doses were superior to lower doses, and the best signal-to-background ratio (SBR) occurred 24 hours after delivery of the ICG.^9,10 For these reasons, 5.0 mg/kg doses are delivered to patients enrolled in human clinical trials the day prior to surgery.

Further work has been performed to evaluate murine models of other types of cancer outside of the brain. In particular, studies of breast cancer, esophageal cancer, prostate cancer, and lung cancer in murine and large animal (canine) models have been performed.^{6,10,11,12,13,14,15,16,17,18,19} These studies demonstrated the capacity of ICG to localize to primary tumors in vivo using SWIG. Further, SWIG has also been found to have the ability to detect residual disease in the wound bed after tumor resection in a mouse and canine model.¹⁸

Our group has also investigated value of SWIG in a rodent model of intracranial brain tumors using the U251-Luc-GFP cell line (not yet published). Two doses were administered, and although peak SBR appears to be seen just an hour after administration, there is a long plateau period from 6 to 72 hours. In this plateau, the SBR remains relatively stable, thus allowing for a broad time window for visualization. Our human clinical studies with SWIG have been performed 24 hours after systemic administration of ICG.

12.3 Peer-Reviewed Studies of SWIG in Other Body Systems

SWIG has successfully been used to identify tumors in multiple other cancer types in humans. Keating et al,16 Newton et al,¹⁹ and Okusanya et al²⁰ showed SWIG could be used to identify pulmonary nodules and various lung cancers. Xia et al showed SWIG could be used to identify metastatic lymph nodes from seminoma and prostate cancer.^10,21 Keating et al showed SWIG could be used to detect thymoma, breast cancer, and lung metastases.^14,15,17

All of these studies demonstrated ICG accumulation in these various tumor types when administered with the SWIG technique. However, one of the major limitations to SWIG in these settings is background fluorescence due to nonspecific accumulation. ICG is delivered heterogeneously to various organs following high-dose delivery due to its presence in the systemic circulation. ICG is also metabolized hepatically, and thus accumulates largely within the liver and bowel. Nonspecific accumulation can lead to potential high background fluorescence in areas near the nodule of interest.

The use of ICG in brain tumors is unique in that the brain has a relatively impermeable blood–brain barrier (BBB) and, additionally, is isolated from these other tissues that can accumulate ICG. Thus, brain tumors are ideal candidates for SWIG, as background fluorescence remains minimal in the intracranial environment. Indeed, in a publication by Lee et al,⁶ the SBR is extremely high in the order of 7 to 10, as the adjacent brain parenchyma reveals minimal fluorescence, as shown in Fig. 12.1 and Fig. 12.2.

12.4 General Findings in Brain Tumors

One of the important findings using SWIG is the ability to localize tumors at depths beyond the view of the unaided eye. Due to the fact that NIR can penetrate brain tissue up to 2 cm, this can permit visualization of brain tumors through the dura, which can aid in the intraoperative planning at the time of the dural opening ( Fig. 12.1c, d). Furthermore, the NIR signal emanating from deep tissues can also assist in planning the corticectomy ( Fig. 12.1e, f; Fig. 12.2c, d). This can help minimize damage to normal adjacent brain structures. This feature is remarkable, as laser excitation of the fluorophore can allow for visualization of the tumor even before the dural opening and before corticectomy ( Fig. 12.2).

SWIG for intraoperative localization does not rely on stereotactic imaging and utilization of surface landmarks, which can be inaccurate due to a variety of factors. One concern is baseline inaccuracy due to shifting of brain surface landmarks during patient positioning and placement of cranial pins. Furthermore, brain shift during tumor resection can render stereotactic navigation unreliable. The real-time intraoperative localization of SWIG has led to further analysis of its ability to study tumor margins.

Fig. 12.1 Second window indocyanine green (SWIG) localizes to glioma. Newly diagnosed glioblastoma. (a) T1-weighted gadolinium-enhanced axial MRI. (b) T1-coronal MRI. (c) Visible light view prior to opening dura. (d) Near-infrared (NIR) signal overlay prior to dural opening. (e) Visible light view upon dural opening. (f) NIR signal overlay after dural opening. Note, the tumor can be visualized through parenchyma. (g) Visible light postresection surgical margin. (h) Residual NIR signal in the resection cavity.

Fig. 12.2 Second window indocyanine green (SWIG) localizes to metastases. Newly diagnosed metastatic adenocarcinoma. (a) Axial T1 MRI. (b) Axial FLAIR MRI. (c) Visible light view through parenchyma. (d) Near-infrared signal overlay through parenchyma. Note, the tumor can be visualized through parenchyma. (e) Visible light view of surgical resection cavity. (f) NIR signal overlay in surgical resection cavity.

In order to study SWIG’s utility in defining surgical margins for tumors, we have compared SWIG to the unaided eye of the surgeon. Each biopsy specimen was coded as either pathologically positive or negative by the senior surgeon using white light only. Immediately after this was recorded, the NIR camera was brought into the field, and the specimen was coded as positive or negative for NIR fluorescence.

It is important to note that in these studies, the extent of resection did not change due to NIR fluorescence findings; biopsies were only taken if deemed safe by the senior surgeon. Using final pathology of the specimen as the gold standard, diagnostic testing for sensitivity, specificity, positive predictive value, and negative predictive value was performed.

12.5 Second Window Indocyanine Green for Brain Tumors

12.5.1 Second Window Indocyanine Green for Glioma

The first published study on SWIG for brain tumors included all gliomas of all WHO grades I to IV.6 Fifteen patients (10 glioblastoma [GBM], 1 WHO grade III anaplastic astrocytoma [AA], 2 WHO grade II astrocytomas, and 1 WHO grade I juvenile pilocytic astrocytoma [JPA]) were enrolled in a phase I clinical trial. On average, NIR fluorescence imaging took place 22.8 hours after infusion of 5.0 mg/kg ICG. The tumors had a mean SBR of 9.5 ± 0.8, and were able to be identified with NIR fluorescence through the dura to a maximum depth of 13 mm.

Of the 15 gliomas analyzed, 12 of them demonstrated positive NIR fluorescence ( Fig. 12.1). Interestingly, the three gliomas that did not demonstrate NIR fluorescence all also did not enhance after gadolinium administration on the preoperative MRI. No association was found between positive fluorescence with tumor size and volume, histology, or injection time.

Surgical margins were analyzed at the completion of the resection ( Fig. 12.1g, h). The NIR imaging device was used to scan the resection cavity and areas of residual fluorescence were biopsied if deemed safe by adjacent anatomy by the senior surgeon. Diagnostic test characteristics on 71 specimens were taken from the 12 patients with positive NIR fluorescence; 51 (71.8%) were characterized as positive for glioma by pathology, while 61 (85.9%) were positive for NIR fluorescence. Using final pathology as the gold standard, the sensitivity, specificity, positive predictive value, and negative predicted values were calculated based on the surgeon’s impression and NIR fluorescence ( Table 12.1).

This study was the first work to demonstrate the practicality of SWIG as a tool for the detection and resection of gliomas. Gadolinium-enhancing tumors were shown to have detectable NIR fluorescence contrast in the operating room that could be visualized in real time during surgery. The sensitivity of NIR fluorescence using this technique was found to be superior to the unaided eye, although this came at the expense of specificity. Although the specificity appears to be low, perhaps due to the nonspecific accumulation of ICG within tumors via the EPR effect, we believe the hematoxylin and eosin (H&E) can oftentimes underestimate the presence of tumor in the biopsied sample, and as such, we believe that the true rate of false positives is lower than presented.22 SWIG has become a promising tool for identifying gliomas both through the dura and through limited brain parenchyma. Its utility for margin detection and determination of extent of resection warrants further investigation.

Table 12.1 Diagnostic characteristics of second window indocyanine green for glioma

	Visible light vs. pathology (%)	Near-infrared positivity vs. pathology (%)
Sensitivity	84.3	98.0
Specificity	80.0	45.0
Positive predictive value	91.5	82.0
Negative predictive value	66.7	90.0

Table 12.2 Diagnostic characteristics of second window indocyanine green for all meningioma

	Visible light vs. pathology (%)	Near-infrared positivity vs. pathology (%)
Sensitivity	82.1	96.4
Specificity	100	38.9
Positive predictive value	100	71.1
Negative predictive value	78.3	87.5

12.5.2 Second Window Indocyanine Green for Meningioma

A second published study of SWIG analyzed 18 patients (13 females and 5 males) with meningioma.23 The mean age of the patients was 55 years (range 20–74 years). Eleven patients had convexity meningiomas, 1 had a parasagittal frontal meningioma, and the remaining 6 had intraventricular or skull base meningiomas (olfactory groove, cerebellopontine angle, and medial sphenoid wing). Final pathology revealed 15 WHO grade I meningiomas and 3 WHO grade II meningiomas. The pathological subtypes of these tumors were as follows: 12 meningothelial, 3 transitional, 1 psammomatous, and 2 either lacking or inconclusive of any further pathological grading.

Using the SWIG protocol, NIR fluorescence was able to positively identify 14/18 (78%) meningiomas intraoperatively. The tumors in these patients on average had an SBR of 5.6 ± 1.7. The NIR signal was observable in all 14 of these patients prior to opening the dura ( Fig. 12.3).

Table 12.3 Diagnostic characteristics of second window indocyanine green for WHO grade II atypical meningiomas

	Visible light vs. pathology (%)	Near-infrared positivity vs. pathology (%)
Sensitivity	81.8	90.4
Specificity	100	55.6
Positive predictive value	100	71.4
Negative predictive value	81.8	83.3

The four patients who did not have fluorescent tumors displayed a peculiar “inverse” fluorescence pattern. In these cases, adjacent brain parenchyma had a higher NIR signal than the tumor (SBR = 0.31 ± 0.1). This inversion pattern did not have a correlation with gender, WHO grade, and history of prior surgery or radiation. All of the inversions occurred in women with WHO grade I meningiomas who had no prior surgical or radiation history. Furthermore, maximum tumor diameter, age, presence of peritumoral T2/fluid-attenuated inversion recovery (FLAIR) signal on pre-op MRI, T1 signal intensity, tumor location, BMI, Ki-67, and pathological subtype all did not predict inversion via logistical regression. Only one variable approached statistical significance (p< 0.25) for predicting inversions: time from infusion to imaging. The average time from infusion to tumor visualization in the NIR positive patients was 22.4 hours, whereas in the four inversion patients the average time from infusion to imaging was 24.4 hours. One possible explanation for this is the fact that the nonspecifically and extracellularly accumulated ICG diffuses out of the tumor space into the adjacent brain after an extended period of time.

Interestingly, all four tumors in the fluorescence inversion cases still contained measurable traces of ICG. These tumors demonstrated the inversion pattern in vivo; however, when the tumor was removed and imaged ex vivo, the fluorescent signal was still significantly stronger than the adjacent dura. Thus, these tumors still demonstrated ICG accumulation, and the lack of NIR signal in vivo could also be due to saturation of ICG within the tumor.

Using the previously described mechanism for margin detection, 46 total specimens from the 14 patients with positive tumor fluorescence were analyzed ( Table 12.2). Twenty-three of these were coded as positive for tumor using only white light, and 38 of the specimens were positive for NIR fluorescence. Ultimately, pathology revealed 28 specimens to be positive pathologically for tumor.

An independent analysis was also performed on atypical meningiomas, as this type of tumor is more likely to have residual disease in the margins and has higher recurrence rates. Table 12.3 shows the characteristics for the 20 specimens biopsied from atypical WHO grade II meningiomas.

This study was the first work to demonstrate the practicality of SWIG as a tool for the detection and resection of meningiomas. Using SWIG to determine extent of resection in real time for meningiomas is a very sensitive technique, but comes at the cost of specificity. For higher-grade meningiomas, the specificity improves; however, it is still not as good as a surgeon’s impression using visible light only. Its utility for margin detection warrants further investigation.