Improvement of the sentinel lymph node detection rate of cervical sentinel lymph node biopsy using real-time fluorescence navigation with indocyanine green in head and neck skin cancer

Authors


Correspondence: Yasuhiro Nakamura, M.D., Ph.D., Department of Dermatology, Division of Clinical Medicine, Faculty of Medicine, University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki 305-8575, Japan. Email: ynakamurta3@yahoo.co.jp

Abstract

The standard technique using lymphoscintigraphy, blue dye and a gamma probe has established a reliable method for sentinel node biopsy for skin cancer. However, the detection rate of cervical sentinel lymph nodes (SLN) is generally lower than that of inguinal or axillary SLN because of the complexity of lymphatic drainage in the head and neck region and the “shine-through” phenomenon. Recently, indocyanine green fluorescence imaging has been reported as a new method to detect SLN. We hypothesized that fluorescence navigation with indocyanine green in combination with the standard technique would improve the detection rate of cervical sentinel nodes. We performed cervical sentinel node biopsies using the standard technique in 20 basins of 18 patients (group A) and using fluorescence navigation in combination with the standard technique in 12 basins of 16 patients (group B). The mean number of sentinel nodes was two per basin (range, 1–4) in group A and three per basin (range, 1–5) in group B. The detection rate of sentinel nodes was 83% (29/35) in group A and 95% (36/38) in group B. The false-negative rate was 6% (1/18 patients) in group A and 0% in group B. Fluorescence navigation with indocyanine green may improve the cervical sentinel node detection rate. However, greater collection of data regarding the usefulness of cervical sentinel node biopsy using indocyanine green is necessary.

Introduction

Sentinel lymph node biopsy (SLNB) is now accepted mainly in the management of cutaneous melanoma. SLNB provides prognostic information and identifies patients with nodal metastases whose survival can be prolonged by immediate lymph node dissection.[1] SLNB has also been used for evaluation of regional nodes in patients with non-melanoma skin cancer such as squamous cell carcinoma and Merkel cell carcinoma.[2, 3] SLNB has been performed mainly using two different methods: blue dye injection (dye method) and radioisotope colloid injection (RI method). Combined use of the dye and RI methods is recognized as the standard technique and provides a high sentinel lymph node (SLN) detection rate of over 90%.[4] However, cervical SLNB has particular problems. Lymphatic anatomy in the head and neck is very complex compared with the axillary and inguinal regions. SLN may be small and located in sites that are not easily accessible, for example, in the parotid gland, through which the facial nerve passes. It is sometimes difficult to detect SLN with lymphoscintigraphy because the SLN is often close to the highly radioactive site where the tracer is injected, the so-called “shine-through” phenomenon.

Recently, several authors have reported on the SLNB technique using indocyanine green (ICG) fluorescence imaging (ICG method) in breast and skin cancer patients and showed high SLN detection rates.[5-9] However, these studies involve mainly axillary and inguinal SLNB and include no cervical SLNB or only a small number of cervical SLNB. In this study, we hypothesized that the ICG method in combination with the standard technique would improve the detection rate of cervical SLN.

Methods

Patients

Twenty-six patients who underwent cervical SLNB for head and neck skin cancer between 2004 and 2012 were included in the study and retrospectively reviewed. Patients were excluded from the study if they had distant metastases or clinical evidence of cervical or parotid nodal metastases. We performed cervical SLNB using the standard technique (dye + RI methods; group A) and performed cervical SLNB using the ICG method in combination with the standard technique (group B). All surgical procedures were under general anesthesia and carried out by well-experienced dermatological surgeons who already had experience with at least 30 consecutive cases of SLNB. All the patients signed an informed consent form.

SLNB

In group A and B, lymphoscintigraphy was performed after i.d. injection of 0.4–0.6 mL of technetium-99–tin colloid (Japan Mediphisic, Tokyo, Japan) around the primary site before surgery. Static and dynamic images were taken and each skin site corresponding to the high accumulation of radioisotope was marked with indelible ink. After the induction of general anesthesia, 0.6 mL of 2% isosulfan blue was i.d. injected around the primary site at the beginning of surgery. In group B only, 0.6 mL of 0.5% ICG (Diagnogreen; Dai-ichi Pharmaceutical, Tokyo, Japan) was also i.d. injected simultaneously around the primary site in addition to the procedure in group A as described above. An incision was made over the marked site. All visible blue-staining nodes, hot nodes whose accumulation of radioisotope was one-tenth of the maximum value counted using the hand-held gamma probe (Navigator GPS System; RMD Instruments, Watertown, MA, USA), and fluorescent nodes detected by the near-infrared camera which was constructed by us,[6] were defined as SLN and excised (Fig. 1).

Figure 1.

(a) “X” indicates accumulation of radioisotope in the left submandibular region. A solid straight line indicates incision line. (b) No blue staining is confirmed in the operating field. (c) It is difficult to detect the sentinel lymph node (SLN) by gamma probe because of the “shine-through” phenomenon. (d) The lymphatic flow (arrowheads) and a fluorescent submandibular SLN (arrow) are visible from the facial skin surface and under the mandible through the near-infrared camera.

Histopathological examination

The excised SLN were examined in multiple permanent sections of tissue stained with hematoxylin–eosin and by immunohistochemical analysis with the use of antibodies to S-100 protein, HMB-45 and MART-1 for melanoma, and cytokeratin for squamous cell carcinoma.

Statistical analysis

The detection rates of the SLN in each group were retrospectively evaluated. Fisher's exact test, Pearson's χ2-test and Student's t-test were performed using StatMate IV (ATMS, Tokyo, Japan). A P-value less than 0.05 was considered statistically significant.

Results

Patient characteristics

The patient characteristics of the two groups are shown in Table 1. We performed cervical SLNB on 18 patients in group A and cervical SLNB on 12 patients in group B. The median age of the patients was 67 years (range, 21–82) in group A and 66 years (range, 32–85) in group B. Of these, 19 had melanoma, 10 had squamous cell carcinoma, and 1 had mucoepithelioid carcinoma. The other patient characteristics were also similar.

Table 1. Summary of patient characteristics
CharacteristicGroup A (blue dye + RI), n or median (range)Group B (blue dye + RI + ICG), n or median (range) P
  1. BMI, body mass index; ICG, indocyanine green; malignant melanoma; MEC, mucoepithelioid carcinoma; MM, malignant melanoma; RI, radioisotope; SCC, squamous cell carcinoma.

No. of patients1812
Age, years67 (21–82)66 (32–85)0.42
Sex
Male970.94
Female950.94
Type of skin cancer
MM1090.49
SCC820.24
MEC010.84
Tumor location
Parietal region400.23
Temporal region410.61
Occipital region110.77
Forehead010.84
Eyelid130.32
Cheek200.65
Nose140.13
Ear310.91
Lip110.77
Neck100.41
BMI22 kg/m2 (17–32)22 kg/m2 (16–30)0.80
Diabetes100.43
Steroid use00

Data of cervical SLNB

Table 2 provides the details of cervical SLNB in each group. The total number of SLN was 35 from 20 basins in group A and 38 from 16 basins in group B. There were no SLN which were detected by dye method but not by RI method. In contrast, there were an additional two SLN which were detected only by ICG method from the skin surface but not by dye and RI methods. Bilateral cervical SLN were detected in two patients in group A and four patients in group B. The median number of SLN was two per basin (range, 1–4) in group A and three per basin (range, 1–5) in group B.

Table 2. Details of cervical SLNB
 Group A (blue dye + RI), n (%) or median (range)Group B (blue dye + RI + ICG), n (%) or median (range) P
  1. a

    SLN identified by preoperative lymphoscintigraphy and additionally identified by intraoperative ICG method, which do not coincide with the accumulation of radioisotope. ICG, indocyanine green; SLN, sentinel lymph node; SLNB, sentinel lymph node biopsy; RI, radioisotope.

Total no. of SLNa3538
No. of basins2016
Location of SLN
Level I418<0.05
Level II130.61
Level III221
Level IV101
Level V300.24
Superficial cervical540.71
Parotid/preauricular1380.3
Postauricular430.7
Occipital200.47
Median no. of SLN per basin2 (1–4)3 (1–5)0.32
No. of excised SLN29 (83)36 (95)0.21
SLN status
Micrometastasis10 (33)3 (8)<0.05
Negative20 (67)33 (92)<0.05
Nodal recurrence1 (6)0 (0)0.43

Comparison of detection rates between group A and B

The SLN detection rate in group B was higher than that in group A (group A, 83%; group B, 95%; Table 2). The SLN which could not be excised were four parotid nodes, one level I node and one level II node in group A, and two parotid nodes in group B. Ten of 29 excised SLN had SLN micrometastasis (five melanoma patients and two squamous cell carcinoma patients) in group A and three of 36 excised SLN had micrometastasis (one melanoma patient) in group B (Table 2).

Short-term clinical outcomes and postoperative complications

In group A, three of seven patients with positive SLN underwent selective neck dissection, and three patients underwent radiation therapy; the remaining one patient refused any procedure. A nodal recurrence occurred in one of 12 patients with negative SLN in group A during the follow-up period (18–84 months). In group B, one patient with positive SLN underwent selective neck dissection and adjuvant radiation therapy and there were no patients with nodal recurrence during the follow-up period (3–22 months). No adverse effects related to the ICG method occurred in group B. Only one of 12 patients in group B had transient facial nerve paresis of the marginal mandibular branch, which resolved 8 weeks after surgery. The remaining 11 patients in group B and all the patients in group A had no SLNB-related morbidity such as wound infection, wound breakdown, bleeding, lymphedema, lymphocele or facial nerve paresis.

Discussion

Although several authors have reported a high detection rate in SLNB for head and neck melanoma,[10-12] the detection rate of SLN in combination with the dye method and the RI method in the cervical region is generally less than that in the inguinal or axillary regions. In the MSLT-I trial reported by Morton et al.,[4] the SLN detection rate in the cervical region (84.5%) was clearly lower than that in the inguinal (99.3%) or axillary regions (96.6%). Several authors infer that the lower incidence of successful SLN detection in the neck may reflect disadvantages such as the complex lymphatic anatomy, the small size of lymph nodes in this region and the shine-through phenomenon.[4, 13] In addition, blue-staining SLN often cannot be detected even when blue dye is injected because of the short staining period for blue dye in cervical SLN due to the rapid and complex cervical lymphatic flow.[13, 14] In our study also, over half of the SLN showed no blue-staining.

To overcome these disadvantages described above, we tried performing cervical SLNB using the ICG method in combination with the standard technique and evaluated whether the ICG method improved the detection rate of cervical SLN. ICG is a widely used diagnostic reagent used in various examinations such as cardiac output, hepatic function and retinal angiography.[5] It binds with albumin and generates a peak wavelength of 840-nm near-infrared fluorescence when excited with 765-nm light.[15] Using a near-infrared camera intraoperatively, it is possible to observe the ICG as a subcutaneous lymphatic flow and SLN in the fluorescence images after i.d. injection of ICG around the primary site. Several authors have reported the SLNB technique using the ICG method in breast and skin cancer patients, showing high SLN detection rates.[5-9, 16] However, these studies involve mainly axillary and inguinal SLNB and only a small number of cervical SLNB has been reported.[13, 17] To our knowledge, there are also no comparative studies regarding cervical SLNB using the ICG method.

In our study, the detection rate of cervical SLN in group B (dye + RI + ICG methods) was higher than that in group A (dye + RI methods), although it was statistically not significant. As we have previously reported,[6] the lymphatic flow extending from the primary site to just in front of the SLN could be observed by ICG method in most patients. However, the lymphatic flow and SLN whose locations were anatomically deep were not visible from the overlying skin. In our experience, the visible range with a near-infrared camera is not more than 1 cm beneath the overlying skin. Despite these circumstances, the fluorescent SLN could be observed after skin incision in most patients. By observing the fluorescent SLN, we could detect most SLN even though it is difficult to detect SLN by the RI method because of the shine-through phenomenon (Fig. 1).

The mean and median number of SLN per basin was more in group B than in group A. The small particle size of ICG, with a molecular weight of 774.96, allows a smooth flow along the lymphatic vessels.[7] It may lead to detection of SLN not detectable by the RI method because of poor flow of the radioactive tracer,[7] and may reduce the false-negative rate. Stoffels et al.[8] also reported that two of 11 additional SLN which were only detected using the ICG method showed micrometastasis. Indeed, there were no patients with nodal recurrence in group B in our study. However, the follow-up period after SLNB was relatively short in this study. A longer follow-up period is needed to confirm the false-negative rate.

As for the limitation of ICG method, because the position of the skin incision is basically determined based on the findings of the RI method, we cannot decide the precise position of a deep SLN by the ICG method alone. Therefore, cervical SLN mapping still requires the RI method.

We still could not detect two SLN despite the application of all three methods. One case was that we could not clearly detect a parotid SLN using the hand-held gamma probe because of the low radioactivity level and shine-through phenomenon, and we also could not observe any visible blue staining and fluorescence with the near-infrared camera. Possibly, the ICG did not drain into the SLN with radioactivity, or the SLN was located too deep to be visible from the surface of the operating field. The other case was that two parotid SLN seemed to be close to each other. One parotid SLN was detected by the ICG method and excised. However, the whole operative field was contaminated with ICG leaking from the lymphatic vessels as a result of procedures during the first SLN excision. We could not successfully detect the other parotid SLN with a hand-held gamma probe because of the low level of radioactivity and also could not confirm any visible blue staining. ICG shows very high fluorescence intensity and just a small leak of ICG can easily contaminate the operative field.[6] Thus, it is difficult to detect the remaining SLN after SLNB in the same operative field, which may be the major drawback of the ICG method.

Despite such limitations, the ICG method in combination with the dye and RI methods can be a useful additional option for cervical SLNB in head and neck skin cancer and improve the cervical SLN detection rate. We especially propose the synergistic use of the ICG method when SLN detection by the RI method seems to be difficult because of the shine-through phenomenon. However, this was a retrospective cohort study and the number of patients was small and the follow-up period was relatively short. Greater collection of data regarding the usefulness of cervical SLNB using the ICG method is necessary.

Acknowledgments

This work was partly supported by the National Cancer Center Research and Development Fund (23-A-22). We thank Brian Purdue of the Medical English Communications Center, University of Tsukuba, for expert English revision.

Ancillary