Pancreatic cancer is a highly fatal malignancy with an average 5-year survival rate of below 5% [1
]. This is largely due to its typically asymptomatic onset, resulting in most patients being diagnosed in late stages. Surgical resection combined with systemic treatment offers the only chance of possible cure. However, nearly 80% of patients present with unresectable disease at diagnosis due to extensive vascular tumor infiltration or distant metastases. During exploratory laparotomy, up to 38% of these patients already have occult metastases or an unresectable primary tumor [2
]. Staging laparoscopies have been advised prior to laparotomies to spare patients from major surgery; but, despite careful patient selection for surgical resection predominantly by computed tomography (CT), metastases still are identified intra-operatively. Moreover, resection with tumor-positive margins (R1) still occurs in up to 50% of patients [3
], leading to postoperative emergence of distant metastases and high recurrence rates [4
]. Neoadjuvant therapy is increasingly being implemented to improve oncological outcomes and increase complete resection rates; however, CT scans struggle to evaluate vascular tumor involvement and to differentiate between viable cancer and therapy-induced fibrosis and necrosis [6
]. This not only limits the prediction of resectability but poses challenges for surgeons while operating [8
]. The ability to distinguish between tumor-positive and tumor-negative tissue is critical for successful oncologic surgery and microscopically radical resection (R0). Fortunately, in recent decades, novel techniques and advanced equipment have emerged that may improve the visualization of anatomical structures and completeness of resections.
One such technique is intraoperative near-infrared fluorescence (NIRF) imaging, also known as fluorescence-guided surgery (FGS). FGS enhances the identification of cancerous lesions, metastatic spread and major vascular structures, the evaluation of blood perfusion, the dissection of appropriate lymph node basins, and the attainment of tumor-free margins during solid cancer surgery [9
]. FGS uses contrast agents with fluorescent characteristics in the NIR region (λ = 700–900 nm), which are visualized by NIR camera systems and displayed on a monitor in real time to the surgeon. FGS can be either targeted or non-targeted. Targeted FGS uses tumor-targeted NIRF probes containing targeting moieties—such as antibodies, peptides, or ligands—which bind with high affinity to proteins or receptors overexpressed on tumor cells and absent on adjacent normal cells. These targeting moieties are conjugated to a fluorophore that emits light in the NIR region [10
]. An ideal NIRF probe would bind with high selectivity to all pancreatic cancer cells, including affected lymph nodes and distant metastases, but not to normal, inflamed, necrotic, or fibrotic tissue cells [11
Non-targeted FGS mostly employs indocyanine green (ICG), which is currently the most commonly used fluorophore. Its absorption peak lies between 790 and 810 nm and emits fluorescence between 820 and 830 nm. Since its approval for clinical use in 1956, ICG has been applied in various branches of surgery and is now routinely used for angiographies, determining cardiac output, assessing hepatic function, and evaluating liver and gastric blood flow. Since the emergence of FGS, the applications of ICG have expanded to mapping sentinel lymph nodes and/or tumors during surgery for breast, lung, liver, and colon cancer, to identifying anastomotic leaks, and to assessing visceral perfusion. Its use has also expanded to endoscopic liver and biliary surgery and has shown potential during staging laparoscopies to detect radiologically occult metastases [2
]. Despite the increasing use of ICG for a variety of applications, there remains a lack of consistency in terms of dose and administration time [15
]. Additionally, its effectiveness during pancreatic cancer surgery is limited due to its non-targeted nature [16
]. Although studies have investigated using ICG for targeted FGS for pancreatic cancer tumors, due to its enhanced-permeability and retention (EPR) effect, Hutteman et al. observed no useful tumor localization in seven out of eight patients, with equivalent ICG uptake in the tumor and healthy pancreas, and no EPR effect observed [17
Despite significant efforts, no tumor-targeted NIRF probes have yet been approved for clinical use during pancreatic cancer surgery. Although it is important to investigate the (bio)chemical, pharmacokinetic, and pharmacodynamic aspects of novel fluorescent probes or dyes before clinical translation, it is critical to remember that preclinical research settings do not always accurately represent the surgical setting. Ultimately, the surgeon must benefit from these advancements to enhance patient outcomes. For this reason, a Delphi study was conducted to acquire a better understanding of experience with and attitudes toward FGS for pancreatic cancer, both with and without ICG, and how these might lead to future research recommendations. Our study is unique in that it focuses solely on surgeons’ perspectives, thereby guiding future research efforts targeting clinical effectiveness.
3.1. Expert Panel
Ultimately, 18 of the 38 invited experts participated in the Delphi survey (47%). The final panel included four surgeons from Germany, four from Italy, three from Japan, three from The Netherlands, two from the USA, and one each from South Korea and Sweden. Although two experts had not performed FGS, they were pancreatic cancer surgeons with sufficient knowledge on the topic, and their extensive experience and sizable publications relating to FGS research verified their eligibility. In addition, all final panel members were acknowledged by other experts as being international experts in their field. All participants were employed at an academic hospital specialized in pancreatic cancer surgery; however, FGS was only performed in hospitals of 13 participants. Further characteristics are summarized in Table 1
3.2. Delphi Results
The results for all statements and questions in each round were considered robust (≥80% of eligible participants voting), as the lowest number of votes was 16/18. Consensus of ≥70% was reached for 61 of 76 statements (80.3%): 33 of 62 (53.2%) in the first round, 27 of 43 (62.8%) in the second round, and 1 of 2 (50%) in the third. Statements for which consensus was reached included: 7 of 11 (63.6%) statements in Module I (Table 2
) on patient preparation and ICG contraindications, 12 of 13 (92.3%) statements in Module II (Table 3
) regarding logistics performing FGS for pancreatic cancer, 23 of 30 (76.7%) statements in Module III (Table 4
) regarding benefits and drawbacks of FGS for pancreatic cancer, 11 of 12 (91.6%) statements in Module IV (Table 5
) regarding where fluorescence imaging is needed during pancreatic cancer surgery, and 8 of 10 in Module V (Table 6
) on future research. Among the statements where consensus was reached, consensus ranged from 70.6–100%.
3.3. General Statements Regarding ICG
All experts unanimously agreed that the general use of ICG is safe with very few side effects, though 83.3% felt patients should be asked about possible allergies before ICG administration. Regarding informed consent, 77.8% of experts agreed that the inability to attain informed consent should not be an absolute contraindication to FGS with ICG; however, no consensus was reached regarding the need for informed consent prior to FGS, with or without ICG. With 88.9–100% consensus, all experts agreed that the dose, concentration, and timing of ICG administration are very important. The consensus optimal dose for tumor imaging was ≤5 mg, which should be determined on a mg/kg basis, and the agreed-upon optimum timing of administration was >1 min before surgery.
3.4. ICG Use during Pancreatic Cancer Surgery
The experts unanimously agreed that ICG not being selective for pancreatic cancer is a limitation of FGS, and that research is necessary to determine the optimum dose, concentration, and timing for ICG use during pancreatic cancer surgery. Nonetheless, 88.9% felt a second intravenous dose of ICG could be given intra-operatively to better visualize pancreatic tumors. Aside from (primary) tumor imaging, 83.3% agreed that administering ICG 24 h before surgery might identify hepatic micro-metastases. Almost all (94.12%) agreed that ICG can evaluate blood flow during organ-preserving surgical techniques—such as the Warshaw technique, spleen-preserving distal pancreatectomy (SPDP), and duodenum-preserving pancreatic head resection (DPPHR)—and is advantageous during pancreatic cancer surgery.
3.5. Fluorescence Imaging during Pancreatic Cancer Surgery
Most (88.9%) experts disagreed with the statement that “intraoperative frozen section analysis is insufficient for identifying resection margins, but fluorescence imaging is”, and 77.8% agreed that though intraoperative frozen section analysis is sufficient, precision analysis can be enhanced by integrating FI into the workflow. Regarding benefits and drawbacks, most experts agreed, with >80% consensus, that fluorescence imaging is useful when visual inspection and palpation are limited, that there are no disadvantages to its use during pancreatic cancer surgery, that it improves intra-operative visualization and is of added benefit during pancreatic cancer surgery, and that real-time flow assessments help to avoid confirmation bias. Between 70–80% agreed that FI (including its equipment) does not interfere with surgical workflow, is easy to use, and neither increases nor decreases the rate of complications, and 72.2% disagreed that FI is unable to distinguish between viable tumor tissue and neoadjuvant therapy-induced necrosis/fibrosis. On the other hand, 94.4% agreed that FI has limited penetration depth, 77.8% that inadequate empirical evidence supporting its efficacy is a major barrier to adopting FI during pancreatic cancer surgery, 72.2% that one limitation is the false positive/false negative fluorescence that may result depending on the distance between the tip of the camera and target tissue, and that FGS still being experimental is a limitation. Lastly, 83.3% agreed that another limitation of FGS for pancreatic cancer is that different pancreatic tumors, such as pancreatic ductal adenocarcinoma (PDAC) versus pancreatic neuroendocrine tumors (panNET), may have different fluorescent features.
3.6. Where Fluorescence Is Needed and Future Recommendations
Module IV (Table 5
) concerns where our experts felt fluorescent guidance is needed most during pancreatic cancer surgery, with either ICG or tumor-targeted probes. Over 80% agreed that fluorescence is needed to visualize the anatomy of the extra-hepatic bile duct during SMA lateral border dissection and to detect and accurately localize metastatic lesions, to determine accurate resection margins, and to visualize surrounding area structures such as the biliary ducts and lymph nodes. Between 75–80% consensus was reached on incorporating fluorescence for accurately localizing lesions, determining extra-pancreatic spread, visualizing vascular structures such as the SMA or SMV, distinguishing between viable tumor tissue and neoadjuvant therapy-induced necrosis/fibrosis, and determining the viability of anastomoses and surrounding organs (e.g., colon, stomach, spleen).
This Delphi demonstrates a high degree of consensus among experts on the safety and potentials of FGS for pancreatic cancer with ICG and tumor-targeted probes. However, despite perceived benefits, experts agreed that FGS, specifically with tumor-targeted probes, should not yet be implemented into routine use due to insufficient empirical evidence proving its benefit over standard back-table methods, and the absence of clinically available tumor-targeted NIRF probes. This study also identified novel directions for future research, such as developing tumor-targeted probes for primary tumor identification in combination with ICG to visualize vasculature and anatomical structures, including pNETs in future research, and standardizing FGS protocols. Achieving these objectives would likely increase surgeons’ willingness to integrate fluorescence imaging into their standard workflow.
ICG has been investigated extensively for FGS, and its safety and effectiveness have been demonstrated in multiple publications [14
]. This is mirrored in our survey’s findings, which included consensus that ICG is safe and should not be presented as experimental. Although dose, concentration, and timing of ICG administration are critical, all experts agreed that further research is needed to achieve uniformity and determine optimal thresholds. These results are consistent with those of another recently published Delphi survey on the general use of fluorescence imaging and ICG [15
Notwithstanding the insights gained from this study, our Delphi’s fundamental limitations are its subjective nature and limited sample size. However, though the results of a Delphi do not reveal “correct” responses, its conclusions do point to truths stemming from the experience of international experts. Although inviting only relevant experts to participate may introduce bias in favor of any subject, the skepticism that can be seen in our results demonstrates experts’ ability to remain impartial. Furthermore, though the size of a Delphi panel can be as small as three members [18
], a larger expert panel could have improved both the accuracy and generalizability of consensus. Still, the experts in this Delphi were geographically diverse, well published, and highly experienced. Similarly, though two experts have not performed any surgeries with fluorescence, their credibility and experience in this field supports their eligibility.
Although ICG has been observed to accumulate in some solid tumors through the EPR effect [23
], pancreatic tumors exhibit minimal to no EPR effect due to their highly stromal nature [24
], and studies investigating the role of ICG for pancreatic cancer imaging have failed to document any useful tumor demarcation [17
]. Similarly, our experts acknowledged that ICG’s lack of specificity for pancreatic cancer is a clear limitation of FGS. These considerations suggest diverting from the use of ICG to visualize pancreatic tumors and further pursue the development and translation of tumor-targeted probes [25
However, this shift does not apply to all aspects of FGS for pancreatic cancer, as our panel viewed ICG as useful for more than just identifying lesions. It was agreed, with strong consensus, that ICG also aids in evaluating blood flow during organ-preserving surgical techniques such as the Warshaw, SPDP, and DPPHR, and that administering ICG at least 24 h before surgery can assist in identifying liver micro-metastases. The latter is especially important, as liver metastases in pancreatic cancer patients are an important prognostic factor, and preoperative imaging modalities—such as CT and magnetic resonance imaging (MRI)—are relatively insensitive to detecting micro-metastases [27
]. For this purpose, ICG can be administered prior to a staging laparoscopy to identify otherwise-undetected liver metastases [2
]. In addition, visualizing the extra-hepatic bile duct, distinguishing between viable tumor tissue and neoadjuvant therapy-induced necrosis/fibrosis, visualizing surrounding structures and vasculature, determining the viability of anastomoses and surrounding organs, and enhancing visualization during SMA lateral border dissection all were surgical steps that could appreciably benefit from fluorescence, according to our experts. The need for improved visualization during these steps has been mentioned repeatedly in various publications [8
]. Consequently, though FGS with tumor-targeted probes could aid in detecting and resecting viable lesions, ICG could be used simultaneously to visualize surrounding structures and vasculature, confirm perfusion in anastomoses, and detect biliary leaks after reconstruction. This could improve resections and organ preservation and lower morbidity rates.
This Delphi also identified new areas of research to explore. Most experts perceived the different fluorescent features of different pancreatic tumors—such as PDAC versus panNETs—as a limitation of FGS, attributable to the hypervascularity of panNETs. Considering that panNETs only account for 5% of tumors [31
], pre-clinical research has primarily focused on PDAC. However, exact identification of a small panNET amendable to minimally invasive parenchyma-sparing resection (i.e., enucleation) can be challenging. FGS can be of benefit in such circumstances, as already shown with methylene blue and ICG [32
]. Furthermore, the following statement, which was added based on the results of an open-ended question, achieved high consensus: “A limitation of FGS for pancreatic cancer is the false positive/false negative fluorescence that may result depending on the distance between the tip of the camera and target tissue”. Although the position of the camera is known to be important for obtaining the appropriate amount of signal [16
], an abundance of fluorescence cameras are available from major brands, as demonstrated by the numerous camera systems our experts use, each having its own configurations and specifications. Although this is not the only cause of false positive/false negative results, standardized set-up, data acquisition, and reporting might contribute to reducing their occurrence. The various imaging systems used also result in considerable inter- and intra-institution reporting variance and can have critical effects during clinical trials or when seeking regulatory approval. As previously suggested, standardized methodologies would be of paramount importance during both processes [35
], whether used during open, minimally invasive, or robotic surgery.
Not only does FGS require a shift in research focus, but it also requires also increased trust from surgeons. Surprisingly, our experts could not agree on whether FGS should be incorporated into routine use for pancreatic cancer. The current standard for intraoperatively identifying resection margins is through frozen section analysis. However, frozen section analysis is time-consuming and sometimes lacks precision, so the introduction of real-time FI is important [36
]. Despite considerable agreement on the benefits of FGS, surgeons still consider frozen section analysis sufficient for identifying resection margins and feel that integrating FI into the workflow merely could
enhance precision. This could be because FGS is still seen as an emerging technology, which is apparent in Table 1
, as 66.7% of experts have only been performing FGS for less than 5 years. This adds to the argument that, despite numerous efforts, investigators have not yet proven that FGS has any significant advantage over current methods. Fortunately, Module III (Table 4
) highlights the perceived benefits and drawbacks of FGS, the majority of which have already been widely discussed [14
], indicating growing belief in FGS.