Orbital Metastases: A Systematic Review of Clinical Characteristics, Management Strategies, and Treatment Outcomes

Simple Summary Orbital metastases may significantly worsen the functional status of oncological patients, leading to debilitating visual impairments. Surgical resection, orbital exenteration, and complementary therapies may result in heterogeneous clinical outcomes. In this systematic review, we aimed to comprehensively analyze the current literature on orbital metastases, describing clinical and imaging features, available management, and treatment outcomes. We found that most orbital metastases occur at later stages after primary tumors, frequently showing diffuse location within the orbit and rarely invading intracranial structures. Biopsy-only techniques were more frequently preferred in view of the less invasive approaches, but surgical resection and orbital radiotherapy were related to improved clinical outcomes. Although patients with primary breast cancer and patients undergoing resection showed superior prognoses, overall survival rates were generally poor, suggesting the need to better understand orbital metastases’ microenvironments for devising optimal systemic treatment strategies. Abstract Background: Orbital metastases often lead to severe functional impairment. The role of resection, orbital exenteration, and complementary treatments is still debated. We systematically reviewed the literature on orbital metastases. Methods: PubMed, Scopus, Web-of-Science, and Cochrane were searched upon PRISMA guidelines to identify studies on orbital metastases. Clinical characteristics, management strategies, and survival were analyzed. Results: We included 262 studies comprising 873 patients. Median age was 59 years. The most frequent primary tumors were breast (36.3%), melanoma (10.1%), and prostate (8.5%) cancers, with median time interval of 12 months (range, 0–420). The most common symptoms were proptosis (52.3%) and relative-afferent-pupillary-defect (38.7%). Most metastases showed a diffuse location within the orbit (19%), with preferential infiltration of orbital soft tissues (40.2%). In 47 cases (5.4%), tumors extended intracranially. Incisional biopsy (63.7%) was preferred over fine-needle aspiration (10.2%), with partial resection (16.6%) preferred over complete (9.5%). Orbital exenteration was pursued in 26 patients (3%). A total of 305 patients (39.4%) received chemotherapy, and 506 (58%) received orbital radiotherapy. Post-treatment symptom improvement was significantly superior after resection (p = 0.005) and orbital radiotherapy (p = 0.032). Mean follow-up was 14.3 months, and median overall survival was 6 months. Fifteen cases (1.7%) demonstrated recurrence with median local control of six months. Overall survival was statistically increased in patients with breast cancer (p < 0.001) and in patients undergoing resection (p = 0.024) but was not correlated with orbital location (p = 0.174), intracranial extension (p = 0.073), biopsy approach (p = 0.344), extent-of-resection (p = 0.429), or orbital exenteration (p = 0.153). Conclusions: Orbital metastases severely impair patient quality of life. Surgical resection safely provides symptom and survival benefit compared to biopsy, while orbital radiotherapy significantly improves symptoms compared to not receiving radiotherapy.


Introduction
Orbital metastases represent 1-13% of all orbital neoplasms and affect approximately 2-5% of patients with systemic malignancies [1][2][3]. Breast, melanoma, and prostate cancers comprise the prevalent primary tumors, and their incidence is increasing due to improved surveillance, systemic disease control and management of oncological patients [3][4][5]. Orbital metastases can often be detected in those with no previous history of cancer due to their common presenting of symptoms of visual disturbance, thus preceding the diagnosis of primary tumors [4,6]. Common presenting symptoms include diplopia, ocular pain, and vision loss, coupled with globe displacement and palpable orbital masses [7,8]. Metastases frequently appear on imaging as irregular and contrast-enhancing lesions in the anterior orbit involving bones and extraocular muscles [8][9][10].
Treatment strategies depend on the clinical presentation and primary tumor pathology; however, a gold standard for treatment has yet to be defined [3,11]. In poor surgical candidates, orbital irradiation can be utilized to facilitate reduction in tumor volume and symptom relief [12]. Surgical debulking is effective in decreasing mass-effect and improving symptoms but can also lead to serious complications such as permanent visual deficits [13]. Chemotherapy, hormonal therapy, and/or targeted therapy have the benefit of simultaneous control of primary and metastatic lesions [7,9].
There is a limited number of individual studies on orbital metastases, and feasible therapeutic options are still debated [14,15]. In this systematic review, we evaluate the clinical features and management strategies of patients with orbital metastases, focusing on factors related to symptom improvement and survival.

Literature Search
A systematic review was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [16] and registered to PROS-PERO (ID: 293984). PubMed, Scopus, Web of Science, and Cochrane were searched from database inception to 10 May 2021 using the combination of the Boolean operators "OR" and "AND", and the search terms "orbital" and "metastases". Studies were exported to Mendeley; duplicates were removed.

Study Selection
Inclusion and exclusion criteria were set a priori. Studies were included if they (1) involved ≥1 patients aged ≥18 years with histologically confirmed distant metastases originating from solid tumors and affecting periocular intra-orbital structures; (2) reported clinical, management and survival data; (3) were written in English. Studies were excluded if they (1) were reviews, technical notes, or autopsy reports; (2) contained only patients with intraocular/eyelid metastases and/or craniofacial/ocular tumors directly extending into the orbit; (3) did not clearly differentiate data between patients with orbital metastases and patients with intraocular metastases or primary orbital lesions; (4) contained insufficient clinical and management data.
Two authors (C.S. and P.P.) independently screened titles and abstracts of all collected citations and subsequently assessed full texts of articles that met the inclusion criteria. A third author (A.S.H.) settled any disagreements. Eligible articles were included, and references were searched to retrieve additional relevant studies.

Data Extraction
Data were extracted from one author (P.P.) and then independently verified by two additional authors (A.S.H. and O.B.A.). Missing data were either not reported or not differentiable from other data. Data included author, year, study design, age, gender, primary tumor, time interval between primary tumor and orbital metastasis, laterality, orbital localization, tissue infiltration, intracranial extension, symptoms, imaging features, extent-of-surgery, surgery techniques, complementary treatment strategies, radiation protocols (i.e., type, fractionation, total dose), clinical/radiological treatment responses, orbital metastases recurrence, local control (LC), overall survival (OS), and survival status. Extentof-surgical resection was defined as "complete resection" for 100% tumor resection and "partial resection" for <100% tumor resection. Clinical and radiological treatment responses were assessed at available time points or at last follow-up. Radiological responses were evaluated following the proposed RANO criteria, describing post-treatment lesions' volumetric changes in patients undergoing tumor biopsy and chemotherapy and/or radiotherapy: complete response (CR)-complete resolution; partial response (PR)-decreased volume; stable disease (SD)-no volume change; progression (PD)-increased volume [17].

Data Synthesis and Quality Assessment
Primary outcomes of interest were clinical characteristics, management strategies, and survival analysis of orbital metastases. Levels of evidence were assessed upon the 2011 Oxford Centre For Evidence-Based Medicine guidelines [18]. Meta-analysis was precluded because all included articles had levels IV-V of evidence, and hazard ratios could not be deducted. Risk-of-bias was independently appraised by two authors (P.P. and O.B.A.) using the Joanna Briggs Institute checklists for case reports and case series [19,20].

Statistical Analysis
The software SPSS V.25 (IBM Corp, Armonk, New York, NY, USA) was used for all statistical analyses, and a bilateral p-value < 0.05 was considered significant for all tests. Continuous variables are reported as medians or means with ranges and categorical variables reported as frequencies and percentages. Clinical and radiologic treatment responses for patients undergoing biopsy and surgical resection were compared using χ2 and Fisher exact tests. Multivariate logistic regression analyses were conducted to ascertain variables associated with clinical and radiological treatment responses, including all the variables deemed of potential relevance-corresponding to a cut-off of p < 0.05 at the univariate analysis-and retained in the multivariate model with a statistically significant power (p < 0.05). In a similar fashion, a multivariate Cox proportional-hazards model was performed to identify variables correlated with OS. Kaplan-Meier methods and logrank tests were used to compare survival outcomes based on patient-level clinical and treatment characteristics. Figure 1 displays the study selection process. The initial search yielded 4689 articles, of which 262 were finally included in accordance with the pre-determined criteria. Forty were case series (including 651 patients) and 222 were case reports, categorized as evidence level IV and V, respectively (overview and references of all included studies are reported in Table S1). Critical appraisal returned low risk of bias for all included studies (Table S2).

Discussion
In this review, we found that surgical tumor resection significantly increased symptom relief and survival compared to biopsy-only. While orbital radiotherapy may effectively prevent tumor progression in the long term, the role of advanced systemic therapies requires further evaluation.
Although some differences in primary tumors have been reported between studies, probably mirroring the underlying geographic variations of cancer rates, we found that breast cancer (36.3%), melanoma (10.1%) and prostate cancer (8.5%) were the most frequent among the 29 different primary neoplasms included in this review [3][4][5]. Similar prevalence rates have been reported for patients with uveal metastases, likely suggesting common routes of tumoral hematogenous spread and organotrophic seeding to both ocular and orbital structures [21][22][23][24]. The true incidence of orbital metastases from primary tumors may be difficult to ascertain only from clinical series of patients with histopathology reports, paving the way to a detection bias in the data collection for this paper. Cancers with more aggressive disease courses may rapidly metastasize to multiple major organs besides the orbit, severely debilitating patient functional status and thus making histological confirmation of suspected orbital metastases unnecessary. This likely explains the lower rates of primary lung cancers (5.6%) compared to other tumors such as carcinoid (6.6%). Indeed, carcinoids are less common malignancies, but they often show better prognoses and slower disease courses, with rare concurrent systemic metastases [5,8]. We also found that the median time interval between primary tumor diagnosis and the onset of orbital metastases was 12 months, further reflecting the long-lasting process of metastatic seeding into the orbit. In line with the literature on patients with choroidal metastases, the detection of orbital lesions preceded the primary tumor diagnosis in 30.1% of patients, with 37 cases of CUPs (4.2%). CUPs represent metastatic carcinomas with no primary neoplasms identified at diagnostic workup and mostly deriving from older case series [3,[25][26][27][28][29]. More recently, upfront whole-body PET/CT scans are frequently recommended in patients with no history of cancer and clinical suspicion of orbital or intraocular metastases to expedite management and systemic treatments [6,30,31].
As opposed to benign orbital tumors, orbital metastases frequently manifest clinically with an abrupt onset of rapidly progressive symptoms [1,32]. We found that most metastases occurred unilaterally and diffusely infiltrated intra/extraconal orbital soft tissues, causing globe displacement with proptosis, diplopia and impaired eye motility [3][4][5]. The direct compression of the optic nerve, commonly at the orbital apex, additionally led to RAPD and vision decline, with a serious impact on patients' functional status [33,34]. Less frequently, paradoxical enophthalmos of the affected eye resulted from the infiltration of neoplastic cells into the extraocular muscles and retro-bulbar stromal tissues causing desmoplasia, fibrosis and globe retraction [35][36][37]. Some ocular symptoms, such as visual impairments and periocular pain, may share similarities between orbital and choroidal metastases, thus requiring further assessment with orbital imaging [7,38]. This is especially the case in patients with no evident orbital or ocular masses. We noted that some primary cancers tend to infiltrate specific orbital tissues, such as breast cancers localizing within the orbital fat pad due to the local hormonal patterns, and melanomas invading and enlarging the extraocular muscles as seen clearly on MRI scans [39,40]. Similarly, prostate and liver cancers commonly infiltrate the orbital bony structures inducing osteoblastic or osteoclastic reactions, and this is better identified on CT scans [7,41].
We found that both incisional and fine-needle aspiration biopsy showed no benefits in clinical and survival outcomes but were chosen over tumor resection in most patients. On par with choroidal metastases, biopsy of suspected orbital metastases is often required for differential diagnosis in patients with no cancer history or for histomolecular characterization when biopsy of other metastatic sites is less viable [6,42]. When surgery is not deemed feasible, biopsy represents a safe alternative advantageous to initiate systemic therapy; however, tumor resection should be preferred in eligible patients. Indeed, we noted that tumor debulking, regardless of the extent-of-resection, led to significant clinical (p = 0.005), radiological (p = 0.007), and survival (p = 0.004) improvement when compared to biopsy. The mechanism for this result may likely be the prompt decompression with relief of mass effect and decrease in tumor burden with increased effectiveness of complementary therapies [14,39,43]. However, we note that our findings may also be related to the selection bias intrinsic in surgical series on orbital metastases because patients planning to receive surgical resection may be characterized by underlying clinical characteristics and tumor features (e.g., satisfactory performance status and single metastasis) not present in patients deemed preferable to receive less invasive and/or nonsurgical treatments.
Regarding orbital exenteration, we found no survival benefit compared to patients undergoing orbital-preserving complete tumor resection (p = 0.153). Such a finding substantiates once again that survival is more likely affected by systemic disease control than metastasis-directed local therapy. Orbital exenteration is often considered for treating aggressive craniofacial malignancies with orbital infiltration and perineural invasion, effectively reducing rates of local recurrences and re-operations [44,45]. In orbital metastases, the limited benefit of orbital exenteration is likely due to the underlying systemic spreading of tumor cells and related poor prognoses [46,47]. Hence, orbital exenteration may be unnecessary for treating orbital metastases in patients with systemic malignancies, and thus surgery should be intended to preserve acceptable quality of life while avoiding highly disfiguring procedures. However, orbital exenteration may play a role in providing relief of severe orbital pain in patients with already poor or absent pre-operative vision function, as we found in seven cases [3,4,39].
In some cases, orbital metastases may extend intracranially into the anterior and middle cranial fossa, similarly to other craniofacial malignancies such as nasopharyngeal carcinomas. While intracranial extension relates with poorer OS in patients with nasopharyngeal carcinomas, we found no differences in clinical (p = 0.582), radiological (p = 0.306), and survival (p = 0.073) outcomes between cranio-orbital and intra-orbital metastases [48]. We postulate that, contrary to primary craniofacial malignancies, both metastatic entities share similar histomolecular characteristics with comparable responsiveness to the systemic and/or radiation therapies when surgical procedures are unviable [49,50]. Still, a more in-depth evaluation of specific tumor microenvironments is highly encouraged to identify favorable therapeutic targets and support less-invasive treatment strategies specifically for patients with cranio-orbital metastases.
Radiotherapy is a well-established palliative therapeutic option for benign and malignant orbital lesions, as well as for periocular ones [11,12,32,[51][52][53]. In line with previous reports on choroidal metastases, we also found that radiotherapy significantly ameliorates clinical and functional status in patients with orbital metastases (p = 0.032), favoring nonsurgical lesion shrinkage and relief of mass effect [22,54]. Although some cases of radiationinduced cataracts have been reported in earlier studies [55], modern image-guided radiotherapy planning allows the delivery of maximal doses to selected targets [56,57], sparing critical orbital structures and preventing the onset of severe adverse events [55,[58][59][60].
Thus, radiotherapy appears to be safe and effective in the treatment of orbital metastases similarly to choroidal metastases, but the severity of radiation-induced complications might differ due to the unfortunate proximity of choroidal metastases to the macula and lens [54,61]. The usefulness of particle therapy, a type of radiotherapy characterized by a peculiarly beneficial dose distribution to the target (Bragg peak), to maximize the sparing of critical ocular structures with respect to the photon-based radiotherapy needs to be further investigated to evaluate its cost effectiveness [62,63]. However, particle therapy has not a widespread distribution, and classic radiotherapy could be more easily accessible. In such critical anatomical sites, safely delivering a high ablative dose in those tumor layers more distantly located from organs at risk while gradually underdosing the successive ones might have a radiobiological rationale in addition to a potentially successful effect. This approach with partial tumor irradiation or with spatially fractionation of the radiation dose is just a hypothesis because it has been effectively tested only for treatment of bulky tumors at different body sites from those examined here [64,65].
While chemotherapy, immunotherapy, and hormonal therapies (for breast and prostate cancers) proved to be significantly effective in concurrently treating both primary tumors and choroidal metastases, their impact on orbital metastases requires further analysis [66][67][68][69].
We found no significant correlation between systemic therapies and clinical (p = 0.247) or survival (p = 0.361) response, perhaps due to the different entry routes of medications into orbital tumors as compared to choroidal tumors [1,28].
The prognosis of patients with orbital metastases is poor with one-year survival rates, (40%) lower than patients with choroidal metastases (52%) and comparable to patients with brain metastases (39-46%), implying that their occurrence is more frequent at the very advanced disease stages, which call for an effective improvement in current systemic therapeutic strategies [22,50,70,71]. In line with previous reports, we found significant survival differences amongst primary tumors (p < 0.001), with one-year survival rates highest in breast cancer (57%) and carcinoid (53%), and lowest in lung (14%), liver (27%), and melanoma (29%) cancers, likely attributed to their underlying aggressiveness and responsiveness to treatments [3,5,8,9,39]. Hence, our findings further substantiate that the current treatments for orbital metastases are palliative and should be aimed at improving symptoms while preserving patient quality of life.

Limitations
Our study has specific limitations. All included articles were retrospective case reports and case series exposed to selection bias, and they comprised a 56-year time-period characterized by major advances in surgical, radiotherapy and systemic oncological treatments, thus likely introducing a chronological bias into our analysis. As mentioned above, our study design involves only histologically confirmed orbital metastases: the exclusion from this data collection of the biopsy-unproven ones could lead to an underestimation of their overall prevalence as well as of their treatment-related outcomes (detection bias). However, our selective inclusion criteria were set to minimize the risks of introducing possible confounding variables related to misdiagnoses of orbital metastases only suspected at clinical and radiological assessments. Indeed, the nature of this systematic review-coupled with the lack of detailed information on criteria used to diagnose suspected orbital metastases among the vast majority of studies found in the literature-prevented us from calculating and adjusting the between-studies inter-observer variability in clinical and/or radiological diagnosis using appropriate statistical tests. The assessment of ocular symptoms was subjective in most studies, based on patient reports. Indeed, we are not able to fully explain why the improvement in symptoms does not correlate with the extent of surgical resection. Presumably, this finding could be affected by an apprehension bias if the risk for disabling and disfiguring consequences discourages patients from submitting themselves to more demolitive surgical procedures. The role of higher aggressive vs. lower palliative radiation doses could not be evaluated due to the lack of granular data across included articles. Due to the commonly missing data in the literature, we could not comprehensively analyze post-surgery adverse events, the impact of tumor size and concurrent systemic metastases on patient survival, or the role of advanced immune/targeted therapies. These confounding variables, coupled with missing information on baseline patients' performance statuses and the likely selection bias related to our inclusion criteria, may have also limited our survival analysis based on surgery vs. biopsy.

Conclusions
Orbital metastases are rare, debilitating lesions in oncological patients. Histopathological examination is recommended to guide complementary therapeutic protocols, but surgery is often challenging. Tumor resection, regardless of its extent, showed improved clinical and survival outcomes over biopsy, but the impact of underlying clinical and tumor characteristics should be still considered on a case-by-case basis before planning surgical strategies. Orbital exenteration appears less useful when used in addition to complete tumor resection because no survival benefit was found. Furthermore, the positive clinical impact of orbital radiotherapy may favor its implementation in patients not eligible to undergo surgery or who underwent a subtotal resection. Future prospective studies are required to better understand the role of multimodal systemic therapeutic strategies in the management of orbital metastases based on primary tumor histopathology.