Introduction

Conjunctival melanoma (CM) is rare but devastating malignancy with high recurrence rate, distant metastasis rate and mortality rate [1]. According to previous case series, the local recurrence rate after therapy at 10 years is more than 60% [2, 3]. Distant and local metastasis occur in up to 50% of cases at 10 years [3]. Metastases typically occur to lymph node (cervical, preauricular, parotid, and submandibular), lungs, liver, skin, brain, and bone through lymphatic drainage and hematogenous spread [4]. The melanoma-specific 10-year mortality is about 28–50% [2, 3, 5, 6].

For early-stage CM, wide local excision with “no-touch” technique followed by cryotherapy is the most common surgical treatment [7]. Non-touch technique is performed under general anaesthesia [8]. For advanced disease, orbital exenteration is a reasonable option [9,10,11]. Orbital exenteration is a radical and disfiguring procedure mostly performed for the management of potentially life-threatening malignancies [12, 13]. Nevertheless, criteria for orbital exenteration of CM have not been specified. A previous study from Paridaens [14] suggested patients with clinical factors like extensive diffuse melanoma, unresectable lesions, involvement of unfavourable sites (palpebral conjunctiva, fornices, eyelid margin and caruncle) or tumours fail to achieve local control to accept an orbital exenteration. Recently, another study reported that orbital exenteration was performed in patients with advanced conjunctival melanoma staged cT3 (clinical T3 stage) [15]. Some of the pathological factors such as tumour thickness have also been identified as potential risk factors of orbital exenteration [16]. However, all clinical and pathological factors including tumour location and its biological behaviour should be taken into consideration when making the surgical plan [17]. At present, no research that incorporated all the clinicopathological factors above has been reported. Also, which factor is more predictive of orbital exenteration remains to be explored.

For CM patients who underwent orbital exenteration, few studies have been performed to evaluate their prognosis due to the rarity of this ocular malignancy. Previous study on eyelid tumours (mainly conjunctival melanoma) has shown that the 1-, 3-, 5-year overall survival of patients who underwent orbital exenteration was 96%, 72% and 60%, respectively [18]. Another study in CM patients who underwent orbital exenteration estimated the 1-, 2-, 5-year overall survival as 96.6%, 75.8%, and 51.5% [15]. Moreover, limited studies have been performed to report the prognostic risk factors as positive surgical margins [18], caruncle involvement and greater tumour thickness [14] for patients with orbital exenteration.

For purposes of this study, firstly, we reviewed the clinicopathological characteristics of patients to comprehensively evaluate the risk factors of orbital exenteration. Moreover, we provided an overview of the prognosis of CM patients who underwent orbital exenteration and found the risk factors for unfavourable prognosis.

Materials and methods

Patients

This study was approved by the Ethics Committee of the Ninth People’s Hospital affiliated to Shanghai Jiao Tong University School of Medicine and was carried out in accordance with the Declaration of Helsinki. Informed consent was obtained from all patients. Consecutive patients who were pathologically diagnosed as CM from January 2000 to September 2021 were included. For patients who underwent exenteration, eyelid sparing surgery was performed for tumours that did not involve the eyelids while total exenteration was performed in the rest. Complete resection with pathologically confirmed tumour-free margin intraoperatively was always applied. The exclusion criteria were a short follow-up period (<6 months) or incomplete data collection. Informed consent was obtained from all patients.

Data collection

Medical records and pathological reports were carefully reviewed. Tumour staging was performed according to the eighth edition of the American Joint Committee on Cancer (AJCC) staging system which was presented in the Supplementary Table. Patients were examined regularly at presentation and during follow-up. Neck palpation, ultrasound (US) and contrast enhanced computed tomography (CT) were used to evaluate nodal metastasis. Regression refers to the degeneration of tumour cells and their surrounding epithelial tissue, replaced by an inflammatory infiltration characterized by lymphocytes. Ulceration is defined as loss of overlying surface epithelium. Distant metastases were identified with imaging techniques, including US, CT, magnetic resonance imaging (MRI) or positron emission tomography-CT (PET-CT). Outcomes including recurrence, metastasis and death were collected. Progression-free survival (PFS) was defined as the time from the date of surgery until a documented disease progression, including recurrence or metastasis. Distant metastasis-free survival (DFS) was defined as the date of surgery until a documented distant metastasis. Disease-specific survival (DSS) was defined as time from date of surgery to date of death related to CM.

Statistical analysis

Univariate and multivariate logistic regression analyses were used for assessing risk factors for exenteration. A Spearman’s correlation was run to determine the relationship between pT stage and cT stage or tumour thickness. Survival curves of PFS, DFS and DSS were estimated by the Kaplan-Meier method, and differences between groups were compared by log-rank tests. Univariate cox regression analyses were performed to study the association between risks factors and survival. A p value less than 0.05 indicated statistical significance. Statistical analyses were conducted in SPSS software (V.22.0, IBM, Armonk, New York, USA).

Results

A total of 79 CM patients presented to Shanghai Ninth People’s Hospital over a period of 21 years were included in this study. The baseline characteristics were listed in Table 1. In total, 53 (67.1%) were male and 26 (32.9%) were female. The mean age at diagnosis was 56 ± 13 years. All tumours were unilateral. Forty (50.6%) left eyes and 39 right eyes were involved. At presentation, the 8th edition of AJCC classification was as follows: T1 in 5 patients (6.3%), T2 in 32 (40.5%) and T3 in 42 (53.2%).

Table 1 Demographic, clinical and pathological characteristics of CM patients.
Full size table

Among the cohort, local recurrence occurred in 36 cases (45.6%); regional metastasis occurred in 16 cases (20.3%); distant metastasis occurred in 30 cases (38.0%); disease specific death occurred in 27 cases (34.2%), after a mean follow-up period of 46 ± 39 months (median 32, range 6–195). Among patients with distant metastases, 12 patients received chemotherapy, 11 of whom died during follow-up; mean survival period after metastasis was 14 ± 11 months (median 14, range 4–37). One patient (Patient No.5 in Table 2 with BRAFV600 mutation) who underwent eye-sparing surgery accepted targeted therapy with a combination of BRAF inhibitor and MEK inhibitor after lung metastasis; the last follow-up showed that the patient still survived with disease 7.5 months after metastasis. Another patient who developed lung metastasis after eye-sparing surgery was treated with anti-PD1 and anti-VEGF; the last follow up showed tumour responded partially to the therapy at 7 months after metastasis.

Table 2 Genetic features of CM patients.
Full size table

Next generation sequencing was performed in 16 patients, including 5 patients who underwent orbital exenteration. Several genetic signatures in CM, such as BRAF, NRAS, NF1, TERT and KIT, were identified. 11 patients were identified to have mutations in these 5 reported genes (Table 2). The commonly reported BRAF V600E mutation was found in 5 patients, two of whom developed distant metastasis and one died of metastatic disease.

Characteristics and prognosis of patients with orbital exenteration were overviewed below. Also, several histopathological and clinical features were recorded (Table 1) and some of the factors were identified to be associated with orbital exenteration and with poor prognosis of patients who underwent orbital exenteration, which were discussed in detail in the following text.

Overview of orbital exenteration: clinical features and prognosis

Of the 79 patients, 32 patients (40.9%) required orbital exenteration at presentation or during the follow-up and 6 patients (18.8%) had undergone previous surgical treatments before orbital exenteration. Patients’ age, gender, and laterality distribution were compared in Table 1. Among the 32 patients receiving exenteration, 3 patients were staged cT2 and the rest 29 patients were staged cT3. All three patients with CM staged cT2 underwent orbital exenteration for multiple recurrence.

During a mean follow-up period of 40 ± 32 months (median 34, range 6–137) after exenteration, disease progression occurred in 14 patients (43.7%), including the occurrence of local recurrence (n = 3, 9.4%), regional metastasis (n = 6, 18.8%) and distant metastasis (n = 16, 50.0%). Among 3 patients who developed local recurrence, 2 patients staged cT3c experienced eyelid sparing surgery. The recurrence time was 6 months and 40 months after surgery, respectively. Another patient staged cT3b experienced total exenteration and recurred in the nasopharynx 17 months after surgery. By Kaplan-Meier survival estimates, the 1-, 5-, and 10-year PFS were 77.4%, 34.8%, and 11.6%, respectively. Distant metastases were noted in 16 cases (46.9%), which were in the liver (n = 6), brain (n = 5), lung (n = 4), bone (n = 3) and skin (n = 2). The mean duration from orbital exenteration to distant metastasis was 33 ± 32 months; 7 patients (46.7%) developed distant metastasis within 1 year and 10 (66.7%) developed distant metastasis within 2 years. By Kaplan-Meier survival estimates, the 1-, 5-, and 10-year distant metastasis rate were 22.8%, 56.3%, 67.3%, respectively. During follow-up, 15 (46.9%) patients died of metastatic disease. The 1-, 5-, and 10-year disease related mortality after exenteration were 13.0%, 58.3%, and 79.1%.

Palpebral conjunctiva involvement

Palpebral conjunctiva involvement was significantly associated with exenteration (OR, 4.26; 95% CI,1.63–11.17; P < 0.01) when estimated by univariate analysis. Of the 31 patients with palpebral conjunctiva involvement, 19 patients (61.3%) underwent orbital exenteration. However, multivariate analysis showed no significance (OR, 0.70; 95% CI, 0.17–2.84; P = 0.62).

Among 32 exenterated patients, palpebral conjunctiva (n = 19, 59.4%) was the most common location of the lesions followed by caruncle (n = 9, 28.1%) and forniceal conjunctiva (n = 10, 31.1%). Patients with palpebral conjunctiva involvement has significantly higher risk of disease progression (HR, 4.78; CI, 1.55–14.74; P < 0.01; Fig. 1A), distant metastasis (HR, 3.57; CI, 1.13–11.23; P = 0.03; Fig. 2A) and disease associated mortality (HR, 4.38; CI, 1.23–15.62; P = 0.02; Fig. 3A) in orbital exenteration group.

Fig. 1: Progression-free survival of exenterated CM patients.
figure 1

Kaplan-Meier survival curves by palpebral conjunctiva involvement (A), caruncle involvement (B), histological ulceration (C) and regression (D).

Full size image
Fig. 2: Distant metastasis-free survival of exenterated CM patients.
figure 2

Kaplan-Meier survival curves for by palpebral conjunctiva involvement (A), caruncle involvement (B), histological ulceration (C) and regression (D).

Full size image
Fig. 3: Disease specific survival of exenterated CM patients.
figure 3

Kaplan-Meier survival curves by palpebral conjunctiva involvement (A), caruncle involvement (B), histological ulceration (C) and regression (D).

Full size image

Caruncle involvement

Caruncle involvement was not a risk factor for exenteration (OR, 1.68; 95% CI, 0.61–4.68; P = 0.32), but in 32 patients with orbital exenteration, caruncle involvement predicts poor outcomes. Ten out of 20 patients with caruncle involvement underwent an orbital exenteration by the end of follow-up. Among ten exenteration cases, 1 (10%) developed local recurrence, 3 (30%) developed reginal metastasis, 7 (70%) developed distant metastasis and suffered melanoma-related death.

In patients who underwent orbital exenteration, caruncle involvement was significantly associated with a higher risk of disease-specific mortality (HR, 3.73; CI, 1.25–11.14; P = 0.02; Fig. 3B). Although the 3-, 5-year disease progression and distant metastasis rate were higher in patients with caruncle involvement than without, there were no significance between patients with or without caruncle involvement as to PFS (HR, 2.50; CI, 0.94–6.65; P = 0.07; Fig. 1B) or DFS (HR, 2.43; CI, 0.89–6.59; P = 0.08; Fig. 2B) in the orbital exenteration group. One case with multifocal lesions extended onto caruncle, palpebral conjunctiva and globe was presented in Supplementary Fig. 1.

Tumour thickness

The mean tumour thickness was 6.3 ± 4.7 mm in orbital exenterated patients and 4.0 ± 2.6 mm in exenteration-free group. Among 27 exenterated patients whose tumour thickness were available, tumours measured less than 2.0 mm was noted in 4 patients (14.8%); tumours measured between 2.0 and 4.0 mm in 5 patients (18.5%); tumours measured between 4.0 and 8.0 mm in 9 patients (33.3%); tumours measured greater than 9.0 mm in 9 patients (33.3%). Among 33 patients in eye-sparing group whose tumour thickness were available, the number (percentage) of patients distributed in these four tumour thickness intervals are 7 (21.2%), 11 (33.3%), 13 (39.4%) and 2 (6.1%), respectively.

pT stage was recorded according to patients’ histological features. Among the cohort, pT stage was positively correlated with cT stage (R = 0.93, P < 0.01) and tumour thickness (R = 0.29, P = 0.03). Greater tumour thickness (OR, 1.27; CI, 1.04–1.55; P = 0.02), together with worse cT stage (OR, 50.75; 95% CI, 5.40–477.07; P < 0.01) were identified as significant risk factors for exenteration in the multivariate analyses. However, patients with greater tumour thickness did not show worse PFS (HR, 1.05; CI, 0.94–1.18; P = 0.36), DFS (HR, 1.05; CI, 0.93–1.18; P = 0.41) and DSS (HR, 1.09; CI, 0.97-1.22; P = 0.15) in patients with orbital exenteration in our cohort.

Ulceration

Ulceration as a histological feature was noted in 11 patients (34.4%) in exenterated patients and 12 patients (25.5%) in the eye-sparing group. Of the 11 patients with orbital exenteration, 2 (18.2%) regional metastases, 8 (72.7%) distant metastases and 7 (63.6%) deaths were recorded. The prognostic analysis of the orbital exenteration group showed that compared with tumours absent of histological ulceration, those present with ulceration had a higher risk of disease progression (HR, 3.85; CI, 1.10–13.44; P = 0.04; Fig. 1C), distant metastasis (HR, 4.83; CI, 1.24–18.84; P = 0.02; Fig. 2C) and mortality (HR, 4.36; CI, 1.07–17.83; P = 0.04; Fig. 3C).

Regression

Histological presence of regression was not a risk factor for exenteration (OR, 0.55; 95% CI, 0.14–2.07; P = 0.37). Of the 32 patients who subjecting to orbital exenteration, regression was noted in 4 patients. Patients in orbital exenteration group present of regression showed worse prognosis of PFS (HR, 5.92; CI, 1.46–23.97; P = 0.01; Fig. 1D), DFS (HR, 6.95; CI, 1.71–28.27; P < 0.01; Fig. 2D) and DSS (HR, 3.56; CI, 0.98–12.89; P = 0.05; Fig. 3D) and all 4 patients presenting regression had disease progression and died of melanoma metastasis within 1.5 to 3.5 years after exenteration.

Discussion

In this retrospective study, we investigated the risk factors for orbital exenteration and evaluated the prognosis of exenterated patients in CM. Our findings showed that patients with CM who presented with worse cT category or greater tumour thickness were at higher risk to receive an orbital exenteration. Among CM patients who have undergone orbital exenteration, palpebral conjunctiva melanoma, caruncle melanoma, histological ulceration and regression were identified to be significantly associated with poor prognosis.

It seems that more factors than what have been reported may be considered in making decision of orbital exenteration. Shields et al. [16] reported that CM patients with clinical features like poor visual acuity, amelanotic tumour or extralimbal tumour location were more likely to receive an orbital exenteration. In addition to that, we found that patients with melanoma staged cT3 or with greater tumour thickness were statistically more likely to experience an orbital exenteration. Although, pathological factors such as tumour infiltrating lymphocytes (TILs), mitotic rates, regression and ulceration which reflects tumour behaviour should also be taken for consideration when a surgical plan is formulated, no pathological factors were found in our study to be associated with the ultimate orbital exenteration.

Whether orbital exenteration could improve the prognosis of CM patients were not clearly identified. Palpebral conjunctiva and caruncle, together with fornices, plica and lid margins were identified as unfavourable locations for CM [19]. Despite that some researchers recommended orbital exenteration for CM in unfavourable locations [14], whether orbital exenteration can improve the prognosis of these patients is still unknown. It is of note that our result showed that patients with caruncle involvement was still at high risk of death although an orbital exenteration was performed, which is consistent with the result of a previous study [14]. What’s more, patients with palpebral conjunctiva melanoma, which has been recognized to be predictive of CM metastasis [20], were also identified to have worse outcomes of disease progression, distant metastasis and melanoma related death following orbital exenteration.

Then should orbital exenteration be performed in high-risk CM patients? Among patients with caruncle or palpebral conjunctiva melanoma, the DFS shows no significance between the exenterated patients and eye-sparing patients (caruncle melanoma: P = 0.70; palpebral conjunctiva melanoma: P = 0.77). Among CM patients staged cT3, the median survival time of exenterated patients were longer (orbital exenteration vs eye-sparing: 34.1, range 6.3-137.4 vs 25.2; range 12.0–59.1; P = 0.54). Also, among caruncle or palpebral conjunctiva melanoma patients staged cT3, the median survival time of exenterated patients were still slightly longer (orbital exenteration vs eye-sparing: 29.0, range 6.3–103.0 vs 24.6; range 12.0–52.0; P = 0.84). Although the results were of no significance and required further validation with larger sample size, for patients staged cT3, orbital exenteration might be better for prolonging patient survival, even when they have caruncle or palpebral conjunctiva involvement. With the desire to alleviate the need for orbital exenteration in the past decade, some researchers have suggested wide surgical resections, followed by customized proton beam radiotherapy (PBRT) to replace orbital exenteration [21]. More research can be carried out to verify its feasibility.

Due to the very limited reports on the prognostic factors of CM patients receiving orbital exenteration, the histopathological features available in 26 exenterated patients were reviewed, among which ulceration and regression were identified as predictive factors for poor PFS, DFS and DSS following orbital exenteration. Tumour ulceration is generally considered to be an indicator of more aggressive cutaneous melanoma, and it is one of the main pathological T-category criterions of cutaneous melanoma [22, 23]. In a study of 26 cases with CM, the presence of histopathologic ulceration was proved to be associated with poor prognosis [24]. Esmaeli et al. [25] suggested that ulceration should be added to the main determinants of pT category in CM. Histological regression in cutaneous melanoma was also considered as a prognostic factor [26, 27], but it was rarely reported in CM patients. Of note, other pathological variables like predominant cell type, mitotic rates, microscopic satellitosis, TIL, and perineural or vascular invasion were proved not to be predictive for poor prognosis of CM patients following orbital exenteration in our study. In this context, our results showed the clinical significance of histopathological ulceration and regression in predicting the prognosis of patients with orbital exenteration. For patients with positive results, eradicating the neoplastic tissue alone may be not enough and close follow-up should be recommended after surgery. Further adjuvant therapies following surgical treatment such as radiotherapy, topical chemotherapy and systemic targeted therapy or immunotherapy may be taken when indicated. Proton radiotherapy was reported to be effective in the management of CM and can act as an alternative to exenteration [28]. BRAF inhibitor has been reported to benefit those CM patients with BRAF mutation [29, 30]. Several case series have demonstrated the advantages of PD-1 inhibitors (pembrolizumab and nivolumab) and CTLA-4 inhibitors (ipilimumab) in the treatment of advanced CM [31, 32]. Two patients accepted targeted therapy or immunotherapy after distant metastasis in our study were still survival with disease. Multi-centre clinical study with larger sample size is needed to validate the efficacy. Locally advanced disease should be discussed in multidisciplinary oncology meetings. It is still unclear whether targeted therapies or immunotherapies should be prescribed in a neoadjuvant or adjuvant fashion following surgery with or without proton therapy.

We are aware it should be cautious when interpreting the results owing to a few limitations. Firstly, not every patient’s pathological data is available, which may interfere with our analysis of the prognostic factors for exenterated patients. In addition, this study only showed single-centre-based statistics. Moreover, the relatively small number of events in each category restricted the implementation of multivariate cox regression analysis.

According to the statistic findings of this retrospective study in our centre, several suggestions are provided here for reference. First and foremost, to avoid orbital exenteration, diagnosis and complete extensive resection while the tumour has not progressed to T3 stage or to a greater tumour thickness are necessary. For patients with palpebral conjunctiva melanoma or caruncle melanoma, orbital exenteration has limited efficacy in preventing the melanoma progression to death. As a result, clinical trials with targeted therapies and/or immunotherapies should be suggested. Finally, we recommend a histopathological examination especially ulceration and regression following an orbital exenteration for CM patients.

Summary

What was known before

  • Orbital exenteration is implemented in CM patients with advanced disease to prevent tumour metastasis and local invasion.

  • Some researchers recommended orbital exenteration for CM in unfavourable locations including palpebral conjunctiva, caruncle, fornices, plica and lid margins.

  • Patients with positive surgical margins, caruncle involvement and greater tumour thickness were found to have poor prognosis following orbital exenteration.

What this study adds

  • CM patients with worse cT category (≥cT3) or greater tumour thickness were more likely to receive an orbital exenteration

  • Palpebral or caruncle involvement, histological presence of ulceration and regression, indicate the poor prognosis of exenterated patients.

  • For metastasized cases, targeted therapies/immunotherapies provide promising outcomes for BRAFV600 mutated/ BRAFV600 wildtype patients.