The literature search yielded 530 papers, 7 of which met the inclusion criteria (Fig. 1). Table 1 presents the key study design elements of all the included articles.11,14,15,16,17,18,19 The reports, published between 1999 and 2019, encompassed a total of 344 patients, with cohorts ranging in size from 17 to 105 patients. A follow-up period was unavailable for one study because it was reported for the entire study population and not specifically for our included patients. Six of seven articles were retrospective, and the remaining article was prospective. Together with the included articles, six patients from our institution were analyzed in the meta-analysis. These patients underwent SLNB for N0 melanoma of the HNCM.

PRISMA flow chart of systematic study selection for SLNB of the parotid and superficial parotidectomy in HNCM patients. SLNB, sentinel lymph node biopsy; HNCM, head and neck cutaneous melanoma

All eight studies used preoperative technetium with intraoperative gamma probe localization to map the lymphatic drainage pattern from the primary tumor. Seven of the eight studies also used blue dye intraoperatively. The pooled results of SLNB showed a successful excision rate of 97 % (95 % CI 0.95–0.99; p < 0.0001) in all eight studies (350 patients). Five of the eight studies reported a mean number of sentinel lymph nodes successfully excised, ranging from 2.3 to 2.94 lymph nodes in four studies, with a mean of 1.3 lymph nodes reported in the remaining study (Figs. 2a and 3a).

Funnel plots of SLNB analysis. Each line represents a separate study. A Succesful SLNB excision. B Positive SLNB result. C Temporary facial nerve injury. D Regional recurrent disease for pN0 SLNB patients. SLNB, sentinel lymph node biopsy

Forrest plots of SLNB analysis. Each dashed line represents a pooled proportion. A Successful SLNB. B Positive SLNB results. C Regional recurrent disease for pN0 SLNB patients. SLNB, sentinel lymph node biopsy

The pooled probability of a positive SLNB result was 16 % (95 % CI 0.12–0.20; p < 0.0001) for all successful SLNB excisions, (reported in all 8 studies, total of 340 patients; Figs. 2b and 3b). All but 4 of 54 patients with positive results underwent subsequent SP with or without neck dissection. For 40 of these patients, only one complication, a temporary facial nerve injury, was reported. The four patients who did not have a subsequent SP were free of disease for the mentioned follow-up period of the studies (3 patients in the Picon et al. 19 study with a median follow-up period of 26 months for entire cohort; 1 patient in the current study with a follow-up period of 16 months).

In the absence of data on the complications experienced by the N0 HNCM patients who underwent immediate SP, we compared data from four weighted meta-analyses of SP. The complications reported were temporary and permanent facial nerve damage and Frey's syndrome, which were absent with the SLNB procedure.

Table 2 presents the key study elements for the articles we analyzed.20,21,22,23 All the articles in the SLNB series reported the incidence of facial nerve injury, with a relative risk for temporary facial nerve damage of 0.124 (95 % CI 0.056–0.273; p < 0.0001). Three events for a total of 334 patients were reported by the eight articles (Fig. 4). Seven of the eight studies reported that no other complications occurred, and the remaining article did not report whether other complications occurred (see the funnel plot in Fig. 2c).

Forest plot comparing SLNB with historic series of SP for temporary facial nerve injury. SLNB, sentinel lymph node biopsy; SP, superficial parotidectomy

No events of permanent facial nerve damage occurred in the SLNB group versus 182 (3.3 %) of 5530 events, in the SP group. The RR for permanent facial nerve damage in the SLNB compared with SP was 0.46 (95 % CI 0.17–1.22; p < 0.0001); Fig. 5).

Forrest plot comparing SLNB with a historic series of SP for permanaent facial nerve injuries. SLNB, sentinel lymph node biopsy; SP, superficial parotidectomy

Disease recurrence to any region had a pooled proportion of 8.3 %, with 24 recurrent events in a total 290 patients (reported in 6 of the 8 articles). Failure of the procedure was defined as the experience of a negative SLNB and a regional recurrence. Of 274 patients reported, 11 (in 7 of the 8 articles) had a regional recurrence during the follow-up period, representing a pooled proportion of 4 % (95 % CI 0.02–0.06; p < 0.0009; Figs. 2d and 3d).

Of 156 patients, 2 died of the disease in the follow-up period reported by five of the eight studies (pooled estimate of 0; 95 % CI 0.000–0.017). The one patient, who died 11 months after diagnosis because of metastases to the brain, had a negative result for four excised sentinel lymph nodes.18 The other patient was from our series. He had a negative SLNB result for the two nodes excised from the parotid, experienced systemic recurrent disease 39 months after surgery, and died 54 months after SLNB.