Introduction

To date, ovarian cancer remains the deadliest gynecological disease in the western world [1, 2]. Due to late presentation of clear symptoms, disease is often diagnosed in an advanced stage, where metastasis to the peritoneal cavity has already taken place [3, 4]. The standard treatment consists of surgical debulking followed by combination chemotherapy. Notably, overall survival is greatly impacted by the efficiency of the debulking surgery, with clear improvements seen for complete debulking over (sub-)optimal debulking [5,6,7,8]. However, due to the highly disseminated nature of late stage ovarian cancer, such complete debulking is a difficult, time-consuming, and intensive procedure [5, 7, 9].

There is strong evidence that intraoperative guidance, by which tissues of interest are marked using a fluorescent contrast agent, can lead to a more precise and more thorough resection of tumor tissue, with ultimately reduced mortality and morbidity rates [7, 10]. The feasibility and benefit of fluorescence-guided debulking surgery to detect an increased number of tumor lesions has been demonstrated by van Dam et al. [11] and Hoogstins et al. [12] using fluorescently labeled folate. Since then, several additional clinical trials have investigated the application of targeted fluorescent tracers in the wide field of surgical oncology and showed that an increase in tumor-to-background signal ratio has a beneficial effect on the sensitivity of tumor detection [13,14,15]. These tracers are often based on antibodies that are already being used for immunotherapy [10].

While such monoclonal antibodies bind their targets with high affinity, their long biological half-life does entail certain disadvantages regarding specificity and the time necessary to attain sufficient contrast [16]. Therefore, nanobody-based tracers are believed to have more attractive characteristics [17]. Nanobodies are the antigen-binding domains of heavy chain-only antibodies found in all members of the Camelid family [18]. As demonstrated by our group and others, their unique attributes in terms of stability, specificity, and pharmacokinetics enable rapid and high-contrast molecular imaging, both for nuclear and optical imaging applications [19,20,21,22,23,24,25].

In this study, an anti-HER2 nanobody, labeled site specifically with the near-infrared (NIR) dye IRDye800CW, and previously validated in terms of biodistribution and specific tumor targeting [20], was used to assess the potential of nanobody-based NIR tracers for fluorescence-guided surgery in a mouse model with disseminated peritoneal tumor lesions, mimicking late stage ovarian cancer [20].

Methods

Tracer Preparation

The preparation and characterization of the anti-HER2 NIR-nanobody 2Rs15dCys-IRDye800CW has been described previously [20]. Briefly, after reduction of 2 mg cysteine-tagged nanobody 2Rs15d with × 180 molar excess of 2-mercaptoethylamine (2-MEA, Acros Organics), the nanobody was incubated for 2 h at 37 °C with × 5 molar excess maleimide-activated IRDye800CW (LI-COR) at pH 6. The NIR-labeled nanobody was subsequently purified by size-exclusion chromatography, and dye-to-protein ratio was determined through UV-VIS spectrophotometry (Nanodrop 2000, Thermo Fisher). A mean of 0.6–0.7 IRDye800CW molecules per nanobody was obtained.

Mouse Model of Intraperitoneally Disseminated Ovarian Cancer

HER2-expressing SKOV3.ip1 Fluc+ cells were kindly provided by Prof. Marc Bracke (UGent, Belgium). These human ovarian cancer cells have been transfected with a pFL4.76 plasmid for luciferase expression and selected for improved intraperitoneal growth [26]. The cells were cultured at 37 °C in 95 % air/5 % CO2 in DMEM medium (Gibco, Invitrogen) with 10 % fetal bovine serum (FBS, Merck Millipore) and antibiotics (penicillin (89 units/ml)/streptomycin (89 μg/ml), Gibco, Invitrogen). Female Crl:NU-Foxn1nu mice were obtained from Charles River (France) and injected intraperitoneally with 0.5 × 106 SKOV3.ip1 Fluc+ cells. Tumors were allowed to grow for 30 days.

Experimental Setup

A schematic overview of the experimental setup is provided in Fig. 1. The aim of the experiment was to generate a proof-of-concept of the utility of nanobody-based tracers in image-guided surgery. All animal study protocols were approved by the Ethical Committee for Animal Experiments of the Vrije Universiteit Brussel. The surgical procedures were performed by a team of two trained oncologic surgeons.

Fig. 1.
figure 1

Schematic representation of experimental setup. Mice bearing intraperitoneal tumor lesions were injected intravenously with IRDye800CW-labeled anti-HER2 nanobody. After 90 min, mice were euthanized and subjected to surgery. In group 1, lesions were removed by a conventional surgical protocol, followed by excision of remaining fluorescent tissue using a NIR fluorescent camera. Group 2 underwent directly fluorescence-guided surgery. Bioluminescence imaging was performed before and after each surgery. Resected tissue was categorized as visualized during conventional surgery or not, fluorescent or not, and bioluminescent positive or negative.

Full size image

2Rs15dCys-IRDye800CW (2 nmol based on dye concentration) was injected intravenously via the tail vein, 90 min before the animals were killed for further surgical procedures. In addition, 10 min before killing, 150 mg/kg d-luciferin (Promega) was injected intraperitoneally enabling bioluminescence imaging (BLI) (Optima, Biospace; 5 min per acquisition) of the exposed abdominal region just before surgery.

Mice were allocated randomly to two experimental groups. In the first group (n = 7), complete debulking was performed according to a conventional surgical protocol based on systematic visual inspection under bright light and palpation, followed by excision of remaining fluorescent tissue as detected with the NIR fluorescent camera FluoBeam800 (Fluoptics) (exposure time per frame 50 ms). The second group of mice (n = 6) underwent surgical debulking, guided directly by continuous video-rate fluorescence imaging (Fluobeam800, Fluoptics, exposure time per frame 50 ms). Finally, the operated animals were imaged again using BLI, as were the resected tissues to assess their tumor status. The latter were then categorized as suspected of malignancy during conventional surgery or not, fluorescent or not, and bioluminescent positive or negative, to enable calculation of true positive and false negative rates, and the percentage of false positive lesions. Resected tissue samples which were not considered as tumor tissue, neither based on visual inspection or fluorescence imaging (e.g., kidneys), were placed under category 4 (“Other”).

From an additional set of tumor-bearing animals (n = 3) injected with 2 nmol NIR nanobody, one observer quantified the fluorescence intensity and 2D surface area of resected tumor lesions after acquisition with the FluoBeam800 (acquisition time 30 ms) in Image J, as well as the non-specific fluorescence signal of relevant intraperitoneal organs and tissues, including liver, muscle, abdominal fat, uterus, ovaries, pancreas, intestines, and spleen.

Data Analysis and Statistical Analysis

Based on the BLI read-out of the resected tissues, true positive rates (TPR or sensitivity) and false negative rates (100-TPR) were calculated. In addition, the percentage of false positives (%FP) resected lesions was determined by dividing the number of BLI-negative resected lesions by the total number of resected lesions × 100.

For the quantification of the percentage of BLI signal remaining in the animal after surgery, compared to before surgery, a correction factor of 0.5 % per minute elapsed after initial BLI acquisition was used to account for the natural decay of the BLI signal following intraperitoneal injection of d-luciferin [27]. Data is expressed as mean ± standard deviation. Variables were tested for homogeneity of variance by a Levene’s test. Comparison between the different groups was performed by one-way ANOVA with Bonferroni correction for multiple comparisons using SPSS Statistics software (version 24.0.0, IBM). A p value ≤ 0.05 was considered significant.

Tumor-to-background ratios (TBR) were calculated ex vivo by dividing the mean fluorescent intensity of tumor lesions to the mean fluorescent intensity of peritoneal organs and tissues (excluding liver and kidneys).

Results

Intraperitoneal tumor inoculation led to the formation of numerous individual tumor implants of different sizes on peritoneal organs and tissues [26]. Debulking in mice cadavers was performed either according to a conventional surgical procedure, followed by fluorescence-aided resection, or directly guided by fluorescence. During fluorescence imaging, the kidneys had to be removed because their high non-specific signal (due to renal clearance of nanobody-based tracers) was interfering with the fluorescent signal of the tumor lesions. Thereafter, fluorescence imaging allowed clear visualization of tumor nodules within the peritoneal cavity, up to submillimeter-sized lesions (Fig. 2, Video S1 in Electronic Supplementary data).

Fig. 2.
figure 2

Representative near-infrared fluorescent images acquired along the image-guided surgery procedure. Fluorescent signal in tumor lesions (indicated by green arrows) is clearly discernible from background signal, including signal in the liver (red arrow). Removal of the kidneys (Kd) was required because of their high fluorescent signal.

Full size image

(MP4 11482 kb)

Average duration of surgery was 20 ± 5 min for conventional surgery, 24 ± 4 min for the ensuing additional fluorescence-aided resection (group 1), and 34 ± 9 min when debulking was performed directly guided by fluorescence imaging (group 2).

Conventional surgery resulted in the resection of 143 tissue specimens for group 1; using subsequent fluorescence imaging, an additional 106 specimens were removed. True positive rates were 59.3 and 95.4 % for conventional and conventional surgery followed by fluorescence imaging, respectively. The majority of the false positive fluorescent tissues could be identified as non-specific fluorescent hot spots in stomach content, fluorescent non-invaded abdominal lymph nodes due to partial paravenous tail vein injections, and very faint and diffuse signal in some adipose tissue near ureters, bladder, and urethra (most probably due to spilling of urine). In group 2 (fluorescence-guided surgery group), a total of 305 tissue specimens were resected, of which 93.0 % were confirmed to be tumor lesions according to BLI (true positive rate). Moreover, fluorescence guidance resulted in a decrease in the percentage of false positively detected lesions from 19.6 to 7.1 %, as compared to conventional surgery (Tables 1 and 2).

Table 1 Overview of total number of excised tissue specimens in groups 1 and 2 per category, according to “suspected of malignancy based on visual inspection under bright light and palpation, or not,” fluorescently labeled or not, and BLI positive or negative (BLI-positive specimens are considered the cancerous lesions)
Full size table
Table 2 Calculated true positive and false negative rates for the different types of surgery, as well as the percentage of false positive lesions
Full size table

To further assess sensitivity, residual tumor tissue after surgery was measured through BLI and compared to the initial tumor burden (Fig. 3a). A significantly smaller proportion of BLI-positive residual implants were left behind in the abdomen following fluorescence imaging (be it directly fluorescence-aided debulking or post-conventional surgery) than when debulking was performed based on visual inspection and palpation only (1.05 ± 0.58 and 0.71 ± 0.50 vs 2.87 ± 1.78 %, respectively, p < 0.05) (Fig. 3b).

Fig. 3.
figure 3

a Representative BLI image of the same animal before and after image-guided surgery. *p < 0.05. b Percentage (%) remaining BLI signal after different surgical steps.

Full size image

The average fluorescence intensity of resected tumor lesions was plotted in function of their size in Fig. 4. A positive relationship between signal intensity and size can be observed. Fluorescent signal of most tumor lesions was higher than background signal in peritoneal organs, with an average TBR of 14.4 ± 8.5, reaching up to 42. Despite the slightly higher background signal of the liver than of other peritoneal organs and tissues (except kidneys) (Fig. 4), even small tumor lesions could be discerned on its surface (Fig. 2).

Fig. 4.
figure 4

Fluorescence intensity of all resected tumor lesions plotted in function of the log2 transformation of their surface area. Mean fluorescent intensity of all tumor lesions (black line) and the liver (red line) is also indicated. Background fluorescence intensity of other peritoneal organs (excluding the kidneys) is indicated as a blue-shaded area covering minimum and maximum mean fluorescence intensity of these organs.

Full size image

Discussion

The use of real-time NIR fluorescence image-guidance stands to revolutionize the surgical treatment of complex and/or locally disseminated cancers [10, 28]. More precise and more efficient intraoperative detection of malignant tissue is expected to improve surgical outcome by preventing under- and overtreatment. Peritoneal carcinomatosis is a cancer type highly amendable for fluorescence-guided surgery [13], as prognosis is greatly dependent on the amount of residual tumor after debulking. In the present study, a mouse model emulating the disease state of intraperitoneally disseminated ovarian cancer was used [26]. The model is also representative for peritoneal metastasized cancer from other origins, e.g., gastro-intestinal cancer [29].

We evaluated a previously characterized specific tumor-targeting nanobody conjugated with IRDye800CW as targeted fluorescent contrast agent. A cysteine-maleimide approach was chosen for conjugation of the NIR dye to the nanobody, as this strategy has demonstrated superior pharmacokinetics to the random labeling strategy, which showed high non-specific uptake and poor tumor contrast [20]. Intravenous administration of the NIR nanobody, 90 min prior to surgery only, resulted in selective labeling and accurate visualization of even submillimeter tumor lesions during debulking surgery. As compared to the conventional surgical procedure performed under bright light, accompanying fluorescence guidance significantly increased the number of resected lesions, and consequently, reduced the amount of residual disease, as determined using BLI as gold standard. Given the targeted nature of the tracer and its rapid clearance from blood and non-targeted tissues, specificity of detection was also significantly increased. However, anatomical referencing is still warranted to prevent resection of tissues and organs that could be falsely interpreted as fluorescent cancerous tissue due to normal pharmacokinetics and clearance via excretory organs of the tracer (e.g., kidneys, bladder, ureter, urethra).

Analogous results in terms of significantly reduced residual disease and increased sensitivity and specificity have been reported for fluorescence-guided surgery using other targeted NIR agents in a variety of orthotopic tumor models, as well as in clinical trials [11, 12, 15, 30,31,32,33,34,35]. For both the reported peptide- and monoclonal antibody-based NIR tracers, TBR typically range from two to five at their respective optimal imaging time points, i.e., a few hours and several days after injection. In our study, a mean TBR of 14.4 was obtained as soon as 90 min post-injection. This early time point is very convenient in a clinical setting, since it allows our tracer to be administered just prior to surgery. Nevertheless, for long-lasting surgery, such as debulking of peritoneal carcinomatosis of ovarian, colorectal or other origin, it is important that the fluorescent signal and TBR remain sufficiently high over the entire surgical procedure. We have previously demonstrated that using this NIR nanobody, specific tumor visualization was feasible for at least 24 h, since signal intensity decreases only slowly over time [20]. Inter-study comparison of TBR can, however, be largely dependent on the camera system used, the dose injected, the selected reference background, and the biomarker targeted, including its expression level. In this regard, the tumor marker HER2 was chosen as target in this study. Although HER2 overexpression in ovarian cancer is associated with a more aggressive disease, HER2 overexpression varies widely between the different subtypes of ovarian cancer [36]. Therefore, this study is more likely to serve as a proof-of-concept of the utility of the nanobodies in image-guided surgery. Pan-tumor biomarkers with high and homogenous expression only in malignant tissue, and clearly delineating the invasive front, would be more appropriate for the purpose. To this end, histopathological investigations are currently ongoing to specifically designate adequate biomarkers for image-guided surgery, in addition to αvβ3 integrin, VEGF, CAIX, uPAR, EpCAM, etc., which are often used today [37,38,39]. An advantage offered by the nanobody technology over peptides or small molecules is the possibility to generically design targeted molecular tracers against any biomarker of interest, with predictable in vivo properties [24, 25].

While the rapid blood clearance of nanobody tracers is considered one of their most attractive properties, the associated renal elimination causes high renal retention of the fluorescent signal [20]. This could affect the ability to detect tumor lesions in the near vicinity of the kidneys. Given the larger dimensions and inter-organ distances in humans, as well as the relatively narrow field of view of many clinical camera systems (including laparoscopic systems), this is expected to be less of an issue in the surgical theater.

The very low hepatobiliary clearance of our tracer enabled precise identification of tumor lesions on the liver surface. Despite the use of a NIR fluorescent dye emitting around 800 nm allowing lesion detection up to a few millimeters of depth [10, 40], deeper located lesions and hidden lesions (such as micro-metastasis in lymph nodes) will remain difficult to reveal. To this end, the development of hybrid-labeled tracers, allowing both intraoperative fluorescence and nuclear detection, has been proposed [28, 41,42,43,44]. Whether this strategy is capable to further increase sensitivity remains to be demonstrated.

Conclusion

The present study demonstrates the potential of specifically targeted NIR nanobodies for highlighting intraperitoneally disseminated tumor lesions with high contrast soon after injection during fluorescence-guided surgery, resulting in increased sensitivity and specificity when compared to the conventional surgical procedure. For ovarian cancer surgery, in particular, this technology is expected to improve patients’ outcome by enabling a more complete cytoreductive surgery, enhancing the effectiveness of ensuing chemotherapy, and decreasing surgical duration and associated side-effects.