Near-infrared fluorescence imaging for sentinel lymph node identification in colon cancer: a prospective single-center study and systematic review with meta-analysis

Colon cancer is one of the most common malignancies in the Western world and the number of early-stage tumors (T1 and T2) identified is expected to increase as a result of the introduction of nationwide screening programs [1]. Lymph node metastases are the strongest predictive factor for patient survival [2]. The low risk of lymph node metastasis in these early-stage tumors makes local excision of the primary tumor an attractive treatment option [3]. However, current treatment by segmental resection with en bloc resection of lymph nodes is unavoidable as long as uncertainties about undetected lymph node metastasis remain. Despite complete segmental resection, up to 20–30% patients with early-stage disease will develop distant metastasis and eventually die from colon cancer [4]. This high recurrence rate in node-negative colon tumors could be the result of understaging due to missed occult tumor cells (e.g., isolated tumor cells or micrometastases) during routine histopathological examination or inadequate lymph node harvesting [5‐8]. Detailed examination of all lymph nodes using serial sectioning and immunohistochemistry (IHC) is desirable. However, these techniques are time consuming and expensive and are, therefore, not appropriate for daily practice. On the other hand, the majority of colon cancer patients without lymph node metastasis are exposed to unnecessary surgery-related morbidity and mortality, currently of 13.5% and 2.0%, respectively [9].

The concept of sentinel lymph node mapping (SLNM) in colon cancer as a staging technique has been described frequently, with variable results [10]. The limited penetration depth and fast migration of current blue dyes, resulting in high false-negative rates, are frequently mentioned as serious drawbacks of SLNM. Near-infrared (NIR) fluorescence imaging for SLNM has several properties that are advantageous for the SLN procedure in colon cancer and has already shown promising results in several types of cancer [11‐13]. The relatively high penetration depth and real-time optical guidance are important benefits in SLN identification since lymph node drainage patterns of colon cancer are unknown and SLNs are generally in an unfavorable location beneath a fatty mesocolon.

Since this technique has increased interest in SLNM in colon cancer, it is important to obtain a broader understanding of the technique and of the factors that influence the success of NIR fluorescence imaging for SLN identification. Therefore, a systematic review and meta-analysis was conducted to provide an overview of the current performance of SLNM with NIR fluorescence imaging in colon cancer, and of the factors influencing its sensitivity, detection rate, negative predictive value and upstaging rates. These outcomes are compared with results and experience of a prospective single-center study performed in our hospital which aimed to determine SLNM accuracy using NIR fluorescence imaging in 30 colon cancer patients.

The concept of SLNM to improve lymph node staging in colon cancer has been investigated extensively [10]. Due to the wide variation in reported outcomes and frequent lower accuracy of results compared to breast cancer and melanoma, overall treatment decision-making based on SLN assessment is still not safe. The location of colonic SLNs in the fatty mesocolon and limited penetration depth of blue dye through fatty adipose tissue are frequently mentioned as serious obstacles to accurate SLNM. NIR imaging using a fluorophore as mapping agent has been proposed as a more effective technique for SLNM in colonic malignancies due to its high tissue penetration of up to 1 cm [29].

In this meta-analysis, we included eight studies that described the use of a fluorophore as a mapping agent for SLNM in colon cancer. We found wide variation between studies regarding the NIR fluorescent SLNM techniques employed, resulting in large differences in reported outcomes. Subgroup analyses were, therefore, performed to identify the technically related factors influencing SLNM accuracy. These technical factors included the tracer used, injection site, in vivo or ex vivo SLNM performance, number of injections and timing between tracer administration and SLN identification. Additionally, tracer composition, dosage and concentrations all varied between studies. Pooled detection rates and negative predictive values both showed acceptable results, at 0.96 (95% CI 0.88–1.0) and 0.81 (95% CI 0.73–0.86), respectively. However, an overall low sensitivity of 0.63 (95% CI 0.51–0.74), accompanied by wide variation (33–85%), was reported among included studies. Although subgroup analysis showed comparable results for SLN accuracy, we believe that several technical factors contributed to the poorer results for SLNM using NIR fluorescence imaging in colon cancer compared to those reported in other types of cancer [30].

The first technical factor which should be validated is site of injection for tracer administration. Although meta-analysis results showed no significant differences after subserosal or submucosal injection of dye, notably outcomes in terms of sensitivity and negative predictive value were found after subserosal injection in our single-center study. According to the current literature, this can be explained by the greater accuracy of injection near the tumor after submucosal injection, which also probably improves uptake by all tumor-draining lymphatic vessels [26, 31‐33]. Another possible explanation is difficulty regarding correct needle positioning and maintaining position during tracer administration. We have personally experienced this problem, especially when using the subserosal injection technique. To improve the accuracy of needle positioning, an injection of 1.0 ml saline in the colon wall was administrated before ICG. The raised `bleb` was helpful in confirming correct positioning of the needle into the colon wall. However, an unfavourable side effect of the injection of saline prior to ICG is the increase of hydrostatic pressure at the injection site. This resulted in dislocation of the needle in several patients and extravasation of dye into the peritoneum, which then made SLN identification impossible in these patients due the high intrinsic background fluorescence in the abdominal cavity. We experienced no spillage of dye using the submucosal injection technique. Only one of the reviewed studies described spillage of dye, which in that case occurred after injection into the subserosal layer, but the authors did not mention this as a cause of false-negative SLN procedures [22]. Independently of the injection technique used, correct needle placement and careful administration of tracer without spillage of dye is crucial to a successful SLN procedure. There appears to be a steep learning curve for accurate tracer administration and we, therefore, recommend that this procedure should only be performed by an experienced surgeon or gastroenterologist.

Tracer administration and SLN identification can be performed in vivo or ex vivo. Advantages of the ex vivo SLN procedure include avoidance of patient exposure to dye, interruption of the surgical procedure, and the relative simplicity of the technique. In addition, ex vivo SLNM is thought to allow a more aggressive dissection of the mesocolon, resulting in improvement of SLN identification [34]. However, ex vivo SLN identification after extraction of the specimen disrupts natural lymphatic pathways. Moreover, an ex vivo approach may exclude SLNs located outside the resection area, whereas an in vivo SLN procedure might have identified aberrant lymph node drainage patterns and thus changed the mesocolonic resection margins [35]. In our study, we did not modify our resection margins, but several reviewed studies identified metastatic SLNs which would not have been included in the standard resection specimen [22, 27, 36]. Furthermore, an in vivo approach can potentially facilitate SLN picking, combined with local excision of the primary tumor when no metastases are found in the SLNs. This treatment approach would dramatically change therapy options for patients with early-staged tumors, and could potentially decrease surgery-related morbidity rates while improving patient survival [3]. It must be emphasized that node picking interrupts the standard oncological approach. Second, a higher number of lymph nodes retrieved are associated with prolonged disease-free survival due to better lymph node staging [37]. These facts underline the need for a highly sensitive SLNM technique before minimization of surgery can be justified in cases where no metastases are found in the SLNs [38].

Optimal timing of tracer administration is another technical factor which needs to be improved. Despite the helpful larger diameter of ICG or possibility of using HSA, the tracer still shows fast migration to higher echelon lymph nodes, resulting in a greater number of fluorescent regional lymph nodes. Designating a large number of fluorescent lymph nodes as SLNs is undesirable since advanced histopathological examination must be applied to all these nodes, which is expensive, time consuming and probably not cost-effective. Another contributing factor is time of harvesting between injection and SLN detection. As shown by the results in Table 2 of the reviewed studies, numbers of assigned SLNs were highest when SLNM was performed more than 15 min after injection [20, 27]. A high number of SLNs should be avoided, as non-SLNs could be mistaken for SLNs and unnecessary costs will be incurred. Therefore, surgeons should attempt to follow lymph flow drainage patterns directly after injection which would overcome prolonged time intervals between tracer administration and SLNM.

A wide variety of dye compositions, concentrations and dosages were employed in the reviewed studies. It is important to note that ICG and IRDye800CW should be seen as distinct tracers, as they have their own specific chemical and physical properties. An ICG dose of around 500 ug has been suggested as optimal for SLNM but this has never been clearly investigated in colon cancer [12, 39‐41]. Humanized-serum albumin is frequently added to ICG and IRDye800CW, which may improve metabolic activity and result in a brighter fluorescent signal. In addition, HSA should increase the hydrodynamic diameter, leading to better retention of dye in the SLN. However, results for fluorescent tracer combined with HSA in SLNM are inconclusive [18, 42, 43, ]. More research to optimize concentration, dosage and composition is necessary to improve the fluorescent signal and to allow differences in signal intensity between the SLN and surrounding tissue to be distinguished.

To improve the NIR fluorescence imaging SLN procedure in colon cancer, patient and tumor- related characteristics should be investigated in addition to technical factors. Currie et al. [21] reported a sensitivity of 33% and argued that this was caused by inclusion of large tumors (> 35 mm) and high number of T3 and T4 tumors, which are associated with more advanced tumor stages. Weixler et al. [20] and Liberale et al. [24] agreed that the inclusion of T3–T4 tumors is a potential cause of high false-negative rates, but not tumor size itself. Tumor stage and size are both frequently mentioned as causes of high false-negative rates in the current literature [44]. It is well-established that higher stage disease with more advanced transmural tumors increases the risk of lymph node metastasis. Second, large longitudinal and transserosal tumors could theoretically destroy efferent lymphatic pathways and may involve adjacent lymphatic drainage patterns, which can increase false-negative rates. It must be emphasized that these factors have never been verified as confounders in large studies [10, 45]. However, since more advanced tumor stages already meet criteria for adjuvant chemotherapy, and treatment of these tumors will not be altered by a SLN procedure, we suggest that future studies should try to include only early-stage tumors.

High BMI and additional mesocolonic adiposity are associated with a decreased sensitivity for SLNM [45, 46]. Although none of the included studies confirmed this association, we experienced technical difficulties during SLNM in patients with a fatty mesocolon. First, it must be emphasized that effective penetration of fluorescence is still limited in fatty mesocolon. Additionally, more fatty tissue requires more dissection, leading to additional risk of disrupting the lymphatic vessels, with leakage of dye into the abdominal cavity as a consequence. As a result, SLN identification becomes more challenging since adhesion of dye to fat compromises the NIR view and fluorescing fat could be mistaken for an SLN [21].

Finally, the pathological assessment of SLNs has a major influence on the accuracy of the procedure. Only advanced histopathological techniques are able to detect occult tumor cells (e.g., isolated tumor cells and micrometastases), although the prognostic significance of these occult tumor cells is still unclear. Nevertheless, it has been suggested that micrometastases are associated with a significant reduction in 5-year survival [5, 8]. Therefore, studies of the SLN procedure should use serial sectioning and IHC for histopathological examination to all SLNs when no lymph node metastasis is found after conventional H&E staining [47].

Limitations of the results presented here include the marked variability in the methods used and the small number of patients included in each study, including our own single-center study.

Evidence regarding SLNM with NIR fluorescence imaging in colon cancer is still limited. Better standardization of the technique will be necessary in future trials. These studies should concentrate on early-stage tumors and focus on tracer composition, injection mode and the improvement of real-time optical guidance to the SLN. Moreover, due to several anatomical and technical difficulties, the SLN procedure in colon cancer seems to be more challenging compared to other types of cancer, and considerable expertise will be required before large patient-related studies can be undertaken to validate SLNM as part of the standard surgical treatment in colon cancer.