The Burden of Histologically Confirmed Uveal Melanoma in Aotearoa-New Zealand: A 21-year Review of the National Cancer Registry

Abstract

Purpose: 

The burden of uveal melanoma (UM) in Aotearoa-New Zealand (NZ), a country with the highest global burden of cutaneous melanoma, is unknown. This first, large-scale study of UM in NZ investigates survival and risks of mortality in histologically confirmed UM.

Methods: 

Deidentified epidemiological data on histologically confirmed UM between January 1, 2000, and December 31, 2020, were extracted from the NZ Cancer Registry. The main outcome measures were patient demographics, tumor characteristics, all-cause versus disease-specific survival, and risks of mortality.

Results: 

Histologically confirmed UM constituted 1.5% (n=703) of all-body site melanomas in NZ (n=47,997). UM predominantly affected Europeans (95%), followed by NZ indigenous Māori (4%), Asians (<1%), and Pacific Peoples (<1%), with no eye or sex predilection. Three hundred eighteen (45%) were deceased at follow-up. Of the deceased, 50% died from UM. The 1-, 5-, and 10-year survival from all-cause mortality was 94%, 68%, and 51%, and disease-specific survival was 97%, 79%, and 71%, respectively. Increasing age at UM diagnosis (>60 y), UM arising from nonspecified sites, and mixed cell UM were associated with an increased risk of disease-specific mortality. No difference in disease-specific mortality was found between sex and ethnicity on multivariate and competing risks analysis.

Conclusions: 

Despite the government-funded public eye care and increasing research and awareness on UM globally, the burden of UM in the 21st century in NZ remains comparable to global studies. We continue to observe an earlier presentation of UM in non-European cohorts, particularly in our Māori population, and further studies on UM in NZ are warranted.

INTRODUCTION

Despite being a relatively rare malignancy per se, uveal melanoma (UM) is the most common primary intraocular cancer in the adult eye¹ and is the second most common type of melanoma after cutaneous melanoma.^2,3 UM incidence worldwide ranges from 5 to 11 cases per million population per year^4,5 in predominantly fair-skinned and light-iris populations in the Northern Hemisphere. UM can be vision-threatening and life-threatening, with ~50% of patients developing metastatic disease.⁶ Metastasis is primarily to the liver, with a median survival of 1 year after documented metastases.^7–9

Prognostic factors for survival in patients with UM include age at presentation, sex, tumor size, location, histopathological, and cytogenetic features.¹⁰ In an age-matched cohort, younger patient age at the time of UM diagnosis has been associated with lower rate of metastasis and better survival compared with older adults (>60 y).^1,11,12 Although some studies have observed no sex differences in rates of metastases and death,^5,13,14 others have observed that male patients carry worse prognosis with higher melanoma-related metastasis and death when compared with females.^15,16 The variability in prognosis seen in different age cohorts and gender can be attributable to variable competing risks such as death from other causes and different life expectancy.^14,17 Unsurprisingly, increasing tumor thickness has been found to increase an individual’s risk of metastasis and death, and the validated American Joint Committee on Cancer (AJCC) staging system (seventh edition) now accurately predicts the risk of metastasis at each stage.^18,19 The AJCC staging system (stage I–IV) and survival prediction have now superseded the previous UM size classification and prognosis based on the landmark Collaborative Ocular Melanoma Study, which was small (1.5–2.4 mm in apical height and 5–16 mm in basal diameter), medium (2.5–10 mm in apical height and ≤16 mm in basal diameter) and large (>10 mm in apical height and >16 mm in basal diameter); however, the nomenclature is still used in many studies today.²⁰ Of all UMs, melanomas located in the iris carry the best prognosis while ciliary body (CB) melanomas carry the worst prognosis,¹⁰ and on histopathology, epithelioid cell type is associated with the poorest survival followed by mixed and spindle cell type melanoma.¹⁰ More recently, cytogenetic features have played an important role in survival prognostication, with worse prognosis seen in tumors expressing chromosome 3 loss or chromosome 8q gain.¹⁰

Aotearoa-New Zealand (NZ) has a public health service similar to but predating the National Health Service (NHS) in the United Kingdom, with government-funded, free access to country-wide ophthalmology health care in addition to more than 1000 community optometrists, serving a population of ~5 million in 2020. However, due to the rare nature of this disease, large-scale epidemiology studies on UM in NZ, a country with extremely high ultraviolet-light exposure and the world’s highest incidence and mortality from cutaneous melanomas,²¹ are limited. Notably, ultraviolet radiation is not a significant risk factor in the development of UM.¹ Indeed, there is only 1 small study (n=16) conducted in 2001 on iris melanoma incidence in the South Island of NZ, which suggested that the incidence was higher than reported elsewhere in the world at that time.²² In light of the limited evidence, the present study addresses a gap in our current understanding of the burden of UM in NZ, as well as provides local geographical data for this rare but important disease. It is important to note that since 1995, iodine-125 plaque brachytherapy has been available for the treatment of small-to-medium-sized melanoma (1995–2003, n=92),²³ and from 2004, the radioisotope was changed to ruthenium-196 (unpublished data from 2004 to 2020, n=196). Another treatment modality that is available in NZ for small-to-medium-sized tumors is stereotactic radiotherapy, and 27 patients in NZ were treated with this modality from 2001 to 2012.²⁴ Tumors in NZ that are treated with plaque brachytherapy or stereotactic radiotherapy do not routinely receive a diagnostic or prognostic biopsy unless there is primary treatment failure or recurrence requiring further treatment. The estimated rate of progression despite treatment with iodine-125 in NZ is 12% and 11% with stereotactic radiotherapy; secondary treatment is typically with an enucleation.^23,24 Therefore, tumor removal, typically by enucleation, is carried out on UMs not amenable to the above treatment or based on a patient’s individual decision irrespective of size. The present study using the NZ Cancer Registry (NZCR) only receives notifications of histologically confirmed UM in NZ, and results should be interpreted with this in mind.

METHODS

A retrospective review of histologically confirmed UM between January 1, 2000, and December 31, 2020, inclusive was conducted using data from the NZCR. The NZCR is maintained by the NZ Ministry of Health and collects data from laboratory reports, reported deaths from the Births, Deaths, and Marriages registry, as well as discharge and health care user information from hospitals through the National Minimum Dataset and National Health Index. Ethics approvals were obtained from Auckland Health Research Ethics Committee (AH22023) and Auckland District Health Board Institution (A+9193). This study adhered to the tenets of the Declaration of Helsinki.

Melanoma cases were identified using the International Classification of Disease (ICD) for Oncology third edition morphology codes 8720 to 8780. Melanoma cases of the eye and adnexa were identified with the ICD 10th edition (ICD-10) code C69—which comprised C690 (conjunctiva), C691 (cornea), C692 (retina), C693 (choroid) and C694 (CB), C695 (lacrimal gland and duct), C696 (orbit), C697 (other specified parts of the eye), C698 (overlapping eye and adnexa), and C699 (unspecified). There was no site-specific code for iris melanoma, and in the NZCR, iris tumors are coded together under C694 CB as per the ICD Oncology third edition guidelines, henceforth referred to as “CB/iris” melanomas. For the purposes of analysis, C692 (retina) was considered as choroidal melanoma as retinal melanoma likely represents a misclassification of choroidal melanoma, as demonstrated in previous studies.^25,26 C699 (unspecified) was also included in our analysis following previous major study groups investigating the epidemiology of UM in Europe, the United States, and NZ’s neighboring country, Australia.^26–28 C697 and C699 were grouped together as not specified/other melanoma for the purpose of analysis. ICD-10 codes C690 and C691, C695, C696, and C698 were excluded as they represented either conjunctival or extraocular melanoma. Primary outcome measures included patient demographics, date of diagnosis based on histopathological confirmation, laterality, morphologic description of melanoma, date of death, and geographical location obtained for analysis. The secondary outcome measure was the incidence of histologically confirmed UM. Morphologic descriptions were reported by NZCR according to the AJCC classification (seventh and eighth edition) that recognizes 3 cell types in UM: spindle cell (composed of both spindle A and spindle B cells), epithelioid cell, and mixed cell type (mixed epithelioid and spindle cells).²⁹

Statistical Methods

Sample characteristics are presented using numbers and percentages for categorical variables, and means and ±SD or median and interquartile range for continuous variables. Differences in age at diagnosis by gender, ethnicity, and melanoma site were assessed using linear regression. Kaplan-Meier analysis was used to estimate median survival times. Univariate and multivariable Cox proportional hazard models were used to evaluate possible differences in survival time from UM diagnosis by age at diagnosis, gender, ethnicity, and UM site. Results are reported using hazard ratios with 95% CIs. Annual incidence rates were calculated for all UM sites, with NZ population estimates obtained from Statistics NZ (https://www.stats.govt.nz/topics/population) for the year 2020. Incidence rates were then directly age-standardized to the World Health Organization World Standard Population Distribution (%), based on the world average population between 2000 and 2025 (https://seer.cancer.gov/stdpopulations/world.who.html), with 95% CIs based on the gamma distribution. Trends in incidence rate across the study period were assessed using linear regression, with the study year as the independent variable. Model assumptions were verified for all analyses, and statistical significance was set as a P-value <0.05. Statistical analyses were conducted using R version 4.1.1.³⁰

RESULTS

Demographics and UM Characteristics

From 2000 to 2020, 47,997 patients in NZ were diagnosed with melanoma from all organ sites. Specifically, 703 histologically confirmed UM cases were identified, which comprised 1.5% of all melanoma cases. 76% of UM arose from the choroid, 18% from the CB or iris, and 6% from nonspecified or other sites. Demographics and UM characteristics are summarized in Table 1.

TABLE 1 - Summary of Demographics and Uveal Melanoma Characteristics for (n=703) Patients Diagnosed in New Zealand Between 2000 and 2020

Variable	Category or units	N (%) or mean±SD	NZ population 2018 (%)*
Sex	Male	351 (50)	49
	Female	352 (50)	51
Ethnicity	European	669 (95)	72
	Māori	25 (4)	16
	Asian	6 (1)	15
	Pacific People	3 (<1)	9
Melanoma site	Choroid (includes n=3 retina)	532 (76)	—
	Ciliary body/iris	129 (18)	—
	Nonspecified or other	42 (6)	—
Morphology	Melanoma, not further specified	305 (43)	—
	Spindle cell	229 (33)	—
	Type A	16 (7% of all spindle)	—
	Type B	141 (62% of all spindle)	—
	Not otherwise specified	72 (31% of all spindle)	—
	Mixed cell	137 (20)	—
	Epithelioid	30 (4)	—
	Nodular (note from nonspecified uveal melanoma sites)	2 (<1)	—
Disease extent	Localized to eye	408 (58)	—
	Invasion of adjacent tissue/organ	32 (5)	—
	Metastasis	25 (4)	—
	Unknown	238 (34)	—
TNM	T category	N (% of total cases reported 2010–2020 n=400)	N (% of total T category n=134)
	pT1	39 (10)	39 (29)
	pT2	47 (12)	47 (35)
	pT3	25 (6)	25 (19)
	pT4	23 (6)	23 (17)
	Not reported	266 (66)	—

^*Results add up to over 100% due to people declaring multiple ethnicities.

NZ indicates New Zealand.

UM was equally distributed between genders and predominantly identified in Europeans (95%), followed by Māori (4%) and <1% in each of Pacific Peoples and Asians. Where laterality was documented (n=462, 66%), UM was equally distributed between both eyes. A detailed morphologic description was provided in 57% (n=398) of UM cases, being predominantly spindle cell, followed by mixed cell and epithelioid cell melanoma. Two thirds of UM cases (n=465, 66.1%) had documented disease extent, the majority reported as localized disease. Only one fifth of cases (n=135, 19%) had an accurately documented T category using the AJCC TNM system. On further analysis, 134 of the 135 (99%) documented T categories were reported from the years 2010 onwards (n=400, 2010–2020) and the remaining one was reported in 2004. We have included the breakdown of the 134 cases with reported T category and secondary percentages at the bottom of Table 1.

The median follow-up time was 5.5 (interquartile range= 2.4–9.7) years. The mean age at diagnosis was 63.6±14.4 years (range=10.2–98.1 y). There was no significant difference in age at diagnosis between sexes, but non-European populations and CB/iris melanoma cases had a statistically lower age at diagnosis as seen in Table 2.

TABLE 2 - Age at Diagnosis of Uveal Melanoma in New Zealand, by Sex, Ethnicity, and Melanoma Site—Results From Individual Linear Regression Models (n=703)

Independent variable	Category	N	Age at diagnosis, in years mean±SD	Mean difference (95% CI)	P
Sex	Male	351	64.5±13.1	Reference	—
	Female	352	62.8±15.6	−1.7 (−3.8, 0.4)	0.119
Ethnicity	European	669	63.9±14.5	Reference	—
	Māori, Asian or Pacific	34	57.8±12.1	−6.1 (−11.1, −1.2)	<0.001*
	Māori	25	58.5±10.0	2.4 (−11.1, 0.4)	0.068
	Asian	6	57.6±19.5	−6.3 (−17.9, 5.2)	0.284
	Pacific Island	3	51.9±15.0	−12.0 (−28.3, 4.3)	0.151
Melanoma site	Choroid	532	64.2±14.0	Reference	—
	Ciliary body/iris	129	60.0±15.6	−4.3 (−7.0, −1.5)	0.002*
	Nonspecified or other	42	66.6±14.4	2.4 (−2.1, 6.9)	0.295

^*Indicates statistical signifcance.

Significance values are in bold.

Primary Outcomes: Mortality and Survival Analysis

The estimated median survival time from diagnosis was 10.3 years (95% CI, 8.7–11.8 y). Of the 703 cases of UM, 318 (45%) were deceased from all causes at the time of follow-up. Of those that were deceased, 158 (50%) died from UM. The 1-, 5- and 10-year survival from all-cause mortality was 94%, 68%, and 51% while the disease-specific survival was 97%, 79%, and 71% (Fig. 1).

FIGURE 1: Kaplan-Meier survival analysis and life tables comparing all-cause and melanoma-specific mortality for patients with uveal melanoma in New Zealand.

On the multiple Cox proportional hazards model (Table 3), increasing age at UM diagnosis from age 50 years and above was associated with an increasing risk of all-cause mortality. Melanoma arising from nonspecified or other sites and mixed cell UM was also associated with an increased risk of all-cause mortality. On univariate analysis, females were associated with a lower risk of mortality, but on multivariate analysis, the difference in mortality risk between sexes was lost.

TABLE 3 - All-Cause Mortality Analysis of Uveal Melanoma Using Cox Proportional Hazards Models (n=701, Including n=318 Deaths)

Variable	N deaths (%)	Group	Univariate HR (95% CI)	P	Multivariable HR (95% CI)	P
Age group	8 (2.5)	<40	Reference	—	Reference	—
	16 (5.0)	40–49	1.61 (0.69–3 .76)	0.272	1.56 (0.66–3.68)	0.307
	51 (16)	50–59	2.66 (1.26–5.60)	0.010*	2.29 (1.08–4.86)	0.032*
	81 (26)	60–69	3.64 (1.76–7.53)	0.001*	3.45 (1.66–7.19)	0.001*
	99 (31)	70–79	6.81 (3.30–14.0)	<0.001*	6.25 (3.00–13.04)	<0.001*
	62 (20)	80+	9.93 (4.74–20.82)	<0.001*	9.37 (4.42–19.85)	<0.001*
Sex	174 (55)	Male	Reference	—	Reference	—
	143 (45%)	Female	0.77 (0.61–0.96)	0.019*	0.82 (0.65–1.03)	0.086
Ethnicity	303 (96)	European	Reference	—	Reference	—
	14 (4.4)	Non-European (Māori, Asian or Pacific)	1.02 (0.6–1.75)	0.941	1.32 (0.77–2.28)	0.316
Melanoma site	241 (76)	Choroid	Reference	—	Reference	—
	51 (16)	Ciliary body/iris	0.71 (0.53–0.97)	0.030*	1.00 (0.75–1.37)	0.999
	25 (7.9)	Nonspecified or other	3.32 (2.19–5.02)	<0.001*	3.17 (2.06–4.88)	<0.001*
Morphology	80 (25)	Spindle cell	Reference	—	Reference	—
	13 (4.1)	Epithelioid cell	1.96 (1.09–3.52)	0.025	1.37 (0.75–2.48)	0.305
	70 (22)	Mixed cell	1.85 (1.34–2.55)	<0.001*	1.63 (1.18–2.26)	0.003*
	154 (49)	Melanoma, not further specified	1.53 (1.17–2.00)	0.002*	1.29 (0.98 −1.70)	0.070

^*Indicates statistical significance. Please note Table 3 does not include n=2 cases identified as nodular melanoma.

Significance values are in bold.

HR indicates hazard ratio.

The disease-specific mortality and subgroup analysis are found in Table 4. On the multivariate Cox proportional hazards model, increasing age at UM diagnosis from 60 years and above was associated with an increased risk of disease-specific mortality. Similar to all-cause mortality, melanoma arising from nonspecified or other sites and epithelioid cell UM were found to carry an increased risk of disease-specific mortality. Of note, no patients with a spindle cell type A morphology died from melanoma-related causes. Note: Due to few numbers of disease-related mortality from remaining spindle cell subgroups, spindle cell B melanoma (n=33) and spindle cell not otherwise specified (n=4) were analyzed together as the reference.

TABLE 4 - Melanoma-Specific Survival Analysis of Uveal Melanoma Using Cox Proportional Hazards Models (n=701, Including n=158 Melanoma-Specific Deaths and n=159 Deaths From Other Causes)

Variable	N deaths (%)	Group	Univariate HR (95% CI)	P	Multivariate HR (95% CI)	P	Competing risks HR	P
Age group	5 (3.2)	<40	Reference	—	Reference	—	—	—
	15 (9.5)	40–49	2.32 (0.84–6.39)	0.103	2.39 (0.86–6.64)	0.095	2.44 (0.90–6.65)	0.081
	33 (21)	50–59	2.65 (1.04–6.79)	0.042*	2.35 (0.91–6.11)	0.079	2.48 (0.97–6.38)	0.059
	43 (27)	60–69	2.78 (1.10–7.02)	0.031*	2.63 (1.03–6.71)	0.043*	2.38 (0.95–6.00)	0.065
	42 (27)	70–79	3.89 (1.54–9.84)	0.004*	3.60 (1.40–9.24)	0.008*	2.85 (1.12–7.24)	0.028*
	20 (13)	80+	4.17 (1.56–11.16)	0.004*	3.65 (1.35–9.87)	0.011*	2.36 (0.86–6.46)	0.095
Sex	81 (51)	Male	Refernce	—	Refernce	—	—	—
	77 (49)	Female	0.89 (0.65–1.22)	0.468	1.07 (0.77–1.48)	0.690	1.13 (0.81–1.57)	0.460
Ethnicity	152 (96)	European	Reference	—	Reference	—	—	—
	6 (3.8)	Non-European (Māori, Asian or Pacific)	0.79 (0.35–1.80)	0.582	0.84 (0.37–1.92)	0.678	0.79 (0.35–1.76)	0.560
Melanoma site	120 (76)	Choroid	Reference	—	Reference	—	—	—
	22 (14)	Ciliary body/iris	0.67 (0.43–1.06)	0.089	0.77 (0.48–1.24)	0.281	0.72 (0.45–1.14)	0.160
	16 (10)	Nonspecified or other	4.11 (2.43–6.96)	<0.001*	3.78 (2.17–6.58)	<0.001*	2.77 (1.52–5.05)	0.001*
Morphology	37 (23)	Spindle cell	Reference	—	Reference	—	—	—
	6 (3.8)	Epithelioid cell	1.90 (0.80–4.50)	0.146	1.62 (0.67–3.89)	0.282	1.57 (0.66–3.73)	0.310
	40 (25)	Mixed cell	2.20 (1.40–3.44)	0.001*	2.06 (1.31–3.24)	0.002*	1.94 (1.25–3.02)	0.003*
	75 (47)	Melanoma, not further specified	1.60 (1.08–2.37)	0.019*	1.40 (0.94–2.09)	0.1.03	1.32 (0.89–1.95)	0.170

^*Indicates statistical significance. Please note Table 4 does not include n=2 cases identified as nodular melanoma.

Significance values are in bold.

HR indicates hazard ratio.

On competing risks analysis in Table 4, only 3 factors were identified with increased risk of mortality: age group 70–79 (compared with age group 0–39), melanoma from nonspecified or other sites (compared with choroidal melanoma), and epithelioid cell type UM (compared with spindle cell UM).

Secondary Outcomes: Incidence Analysis

The overall crude and age-standardized annual incidence of histologically confirmed UM in NZ was 7.58±1.34 and 5.56±1.02 cases per million population per year, respectively. The age-standardized annual incidence of choroidal, CB/iris, and nonspecified or other melanoma were 4.16±1.02, 1.08±0.52, and 0.31±0.37 cases per million population per year, respectively. The standardized UM incidence per million per year by sex was 5.72±1.15 in males and 5.39±1.47 in females, with no significant differences between them (mean difference: 0.33, 95% CI, −0.50 to 1.15, P=0.43). There was no significant trend in UM incidence over the study period [−0.05 cases per million per year (−0.13, 0.03), P=0.178]. However, on subgroup analysis, CB/iris melanoma combined had a small but statistically significant downward trend in incidence (−0.05 cases per million per year [−0.08, −0.01], P=0.008) while no changes were observed in choroidal melanomas [−0.01 cases per million per year (−0.09, 0.07), P=0.762] (Fig. 2). Further time-trend analysis in 5-year periods showed no obvious change in age-standardized annual incidence when comparisons were made using directly standardized rate ratios as seen in Table 5.

FIGURE 2: Uveal melanoma incidence in New Zealand between 2000 and 2020, by tumor location.

TABLE 5 - Time-Trend Incidence Analysis Using 5-year Periods From 2000 To 2020 Using Directly Standardized Rate Ratios

Comparator	Reference	Standard rate (per million)	Rate ratio (95% CI)
—	2000–2004	5.57	Reference
2005–2009	2000–2004	5.91	1.06 (0.83–1.29)
2010–2014	2000–2004	6.16	1.11 (0.88–1.33)
2015–2020	2000–2004	4.75	0.85 (0.63–1.07)

DISCUSSION

To date, most UM studies have been conducted in countries in the Northern Hemisphere,^{4,5,28,31–34} and the current study is the largest whole-population study to explore the burden of UM in NZ, a relatively confined population in the Southern Hemisphere.³⁴ Our demographic analysis confirms that, within the NZ population, UM predominantly affects individuals of European ancestry,³⁵ with no significant gender predilection. It has previously been demonstrated that UM occurs in a 1:15 to 1:50 ratio in Black:White patients,^1,33 and we note a 1:26 ratio of UM in Māori:European patients. We know of only one reported NZ case of choroidal melanoma in Māori,³⁶ with no Māori or Pacific Peoples identified in a cohort of iris melanocytic lesions (n=51) treated in the last 20 years in NZ.³⁷ Interestingly, we observed that the age at diagnosis of UM in ethnic groups in NZ with more pigmented skin and iris color (Māori, Asians, and Pacific Peoples) was younger than in European ethnicities. Elsewhere, lower age at UM presentation has also been noted in Asian, Black, Pacific, and Hispanic populations.^1,33,38–40 Māori are NZ indigenous population that has origins from Southeast Asia, Melanesia, and Polynesia, as discovered via genetic analysis.⁴¹ Given the ancestry, it is likely that the earlier age at diagnosis observed in our Māori population may be related to underlying genetic similarities to the Asian and Pacific cohorts above, and thus similarities in the earlier presentation with UM. The overall infrequency of UM in Māori, Asian, and Pacific Island ethnicities may be related to increased skin and eye pigmentation as a protective factor, but occupation, cultural, or environmental factors may affect earlier presentation.¹

The overall 5-year disease-specific survival of UM in NZ (79%) was comparable to other study groups at ~81%.^26,31,42 In our study, there was limited TNM staging documented in the NZCR available for further survival subanalysis and upon further investigation, it was noted that most categorizing of TNM staging for UM in NZ started in 2010. Of the 134 cases (of 400 cases between 2010 and 2020) with documented T staging, we were able to extrapolate a crude estimate of tumor distribution based on size, which was relatively comparable with other large registry studies in the UK and across Europe (France, Germany, and Finland).^43,44 Unfortunately, despite advancing technology in ophthalmic screening and diagnosis in the 21st century, UM diagnosis in up to a third of patients can still be missed despite having had an ophthalmic examination completed within the past 1 to 2 years and continue to present with more advanced disease.^45–47 The lack of change in survival observed in our study likely reflects the lack of truly effective treatment for metastatic UM at the present time.

Melanomas from nonspecified or other sites, increasing age at diagnosis, and epithelioid cell UM were found to have a statistically significant increased risk of mortality. In the absence of individual pathology reports for review, we postulate that the observed higher risk of mortality in individuals with nonspecified or other UM was likely secondary to larger, more aggressive tumors with invasion of multiple ocular sites, making precise anatomic classification challenging. Extraocular invasion of UM correlates significantly with other adverse, clinical, histologic, and genetic risk factors for metastasis,¹⁷ while larger tumors have been associated with an increased risk of metastasis and mortality.^19,48 Meanwhile, the increasing risk of all-cause and disease-specific mortality with increasing age at diagnosis of UM has been recognized previously¹⁰ and may relate to immunosenescence,⁴⁹ a term describing the progressive decline in immune function that develops with age. When compared with spindle cell UM, which is known to have generally better prognosis,^1,10 we observed increased mortality only in mixed cell UM but not in epithelioid cell UM. The lack of associations in epithelioid cell UM is likely explained by the small sample size of epithelioid cell UM (n=30) in our cohort when compared with mixed cell (n=137) and spindle cell (n=229). It has been well established that increasing number of epithelioid cells is associated with a worse prognosis.⁵⁰

We also report, for the first time, the incidence of histologically confirmed UM in NZ. The reported age-standardized incidence of 5.6±1.0 cases per million NZ population per year is within current global estimates in populations of predominantly European ancestry.^{4,5,31,33,42,51} As astutely observed by Stang et al,²⁸ we do recognize that the current estimated incidence is likely to be an underestimate of the true incidence as an increasing amount of UM cases are being detected earlier and are likely smaller and therefore amenable to treatment with eye-preserving therapies such as with plaque brachytherapy, stereotactic radiotherapy, and proton beam radiotherapy. Indeed, in data extrapolated from published^23,24 and unpublished local audits, we estimate that there is an additional 28% of UM cases that were treated by plaque brachytherapy and stereotactic radiotherapy, treatment modalities that did not receive a histologic diagnosis before treatment. Furthermore, it is not currently a standard practice in NZ to obtain a prognostic tumor sample for eye-preserving therapies to reduce the risk of tumor seeding. Presently, the national cancer registry in NZ does not include UM cases without a histologic diagnosis and therefore these cases need to be further identified through treating clinicians dispersed within the 20 district health boards in the country. In comparison to Australia, a neighboring country to NZ, it was noted that their national cancer registries included clinically confirmed cases of UM when this was notified by treating ophthalmologists, and Vajdic et al⁵² found that only 64% of their cases between 1996 and 1998 were diagnosed with a histological diagnosis. Given the similarities in the available treatment modalities between these 2 countries in the period studied, we postulate that approximately one third of the cases in NZ are diagnosed clinically and therefore not captured by the present study.

The strength of this study is attributed to the long study period using a national cancer database, allowing enough time for meaningful data on outcomes to be collected on a rare cancer such as UM. The use of information from NZCR, a national registry, is comparable to previous methodologies used by other UM study groups,^{4,5,26,31,42,51,53} although NZCR currently only collects information on histologically confirmed UM. Furthermore, complete ascertainment of mortality status was obtained for each patient. Although derived from a large national database, this study is not without its limitations. The documentation of tumor morphology and site in the registry was not always complete and due to the deidentified nature of the data, it was not possible for the investigators to obtain individual pathologic reports to review. Furthermore, the NZCR currently classifies iris melanomas alongside CB melanomas, which are known to have contrasting prognosis. Classically, the location of CB melanomas makes early detection difficult, hence, substantial growth, later presentation, and tumor spread can occur without the patient being aware. In contrast, iris melanomas are typically detected earlier owing to their anterior and easily visualized position, and generally they are smaller in size and have lower biological activity.^10,19 Therefore, the inability to run further site-specific analysis on iris and CB melanomas hindered the individual understanding of both disease processes, although we have very recently commented on iris melanomas in a separate 20-year study in NZ.³⁷ Finally, although we can clearly determine the incidence of histologically proven UM, as previously noted, the overall incidence of UM (ie, histologically proven plus cases treated without microscopic confirmation) is likely to be higher. We recommend that future data collection of UM by the NZCR to include clinically confirmed cases, which can be facilitated by a collaboration with local district health boards carrying out eye-preserving UM therapies and notification by local ocular oncologists and ophthalmologists.

In conclusion, the current large (n=703), 21-year cancer registry–based study presents unique data on the status of UM in NZ, an area of research where local studies are presently limited; and the results add to the current, global understanding of this rare but important, life-threatening disease. The 5-year survival of histologically confirmed UM cases in NZ seems to compare similarly to rates found in the literature. Lastly, further UM studies in NZ are needed to better understand the behavior of this disease in a country.