Circulating cell-free DNA does not harbour a diagnostic benefit in cats with feline diffuse iris melanomas

Abstract

Objectives

Feline diffuse iris melanoma (FDIM) is the most common malignant primary intraocular tumour in cats, with reported metastases rates between 19% and 63%. Currently, the only available diagnostic tool for a tentative diagnosis is histopathological examination of the enucleated eye. Therefore, the veterinary ophthalmologist is often faced with the dilemma of whether to enucleate an oftentimes visual eye or to continue monitoring, with the risk of metastases developing. In the past, cell-free DNA (cfDNA) gained more attention in human medicine, especially in the field of oncology. Prior studies have shown the use of cfDNA as diagnostic or prognostic markers in canine and human cancer patients. Therefore, the aim of this study was to investigate cfDNA concentration and integrity in cats with FDIMs compared with cats with benign iris naevi and without ocular abnormalities.

Methods

cfDNA from plasma of cats with iris melanoma (n = 34), iris naevus (n = 30) and without ocular abnormalities (n = 32) were extracted. Primer and probes for feline amyloid beta precursor protein (APP) and beta actin (ACTB) were designed for amplicons of various lengths and quantitative PCRs of extracted cfDNA were performed to measure cfDNA concentration and integrity of the plasma samples. Differences of cfDNA concentrations and integrity levels between the three groups (iris melanoma, iris naevi and controls) were analysed using the Mann–Whitney U-test.

Results

cfDNA concentration and integrity analysis revealed no significant differences between the cats with iris melanoma, iris naevus or the control group (P >0.01). Cats with metastases showed similar cfDNA concentration and integrity to cats without metastases.

Conclusions and relevance

cfDNA concentration and integrity seem to be insufficient as a diagnostic or prognostic marker in cats with FDIMs.

Introduction

Feline diffuse iris melanoma (FDIM) represents the most common intraocular tumour in cats with approximately 50% of all feline intraocular neoplasia.^1–3 FDIMs may present as a nodular or diffuse accumulation of pigment within the iris. ⁴ This potentially highly malignant tumour has a metastasis rate between 19% and 63% of cases in liver, lung, spleen, lymph nodes and bone.^1,5–7 With a mean age of 11 years, FDIMs are typically diagnosed in middle-aged to older cats.

Although histopathological examination remains the gold standard, currently the only available diagnostic tool for a tentative diagnosis of FDIM is an ophthalmological examination. The only current treatment options are melanoma vaccines, sector iridectomy, diode laser ablation and enucleation of the affected eye globe. Findings that may lead the ophthalmologist to the decision of enucleation include growth of the pigmented areas, involvement of the iridocorneal angle, change of pupil shape or mobility, thickening of iridal tissue and elevation of intraocular pressure (IOP). ⁸

Nevertheless, a careful and complete ophthalmological examination of every cat with changes in iridal pigmentation is important as there are differential diagnoses such as iris naevi, melanosis and iridociliary cysts. ⁹ The biggest dilemma for veterinary ophthalmologists is whether to enucleate an oftentimes visual eye, which may not be affected by a malignant process, or to continue monitoring and risk metastases.

In 1948 Mandel and Metais reported the presence of free nucleic acids in human serum. ¹⁰ Although free nucleic acids were ignored for decades following their discovery, today several vital functions of cell-free DNA (cfDNA) have been determined. cfDNA harbours a great source of biomarkers in conditions such as systemic lupus erythematosus, arthritis, hepatitis and various types of cancer, containing higher levels of cfDNA in their blood than healthy individuals. ¹¹

Although pathophysiological processes responsible for cfDNA origin and circulation are not completely understood, it is assumed that cfDNA originates from normal and abnormal cells as a result of necrosis, apoptosis and secretion.^12,13 It is assumed that tumours induce necrosis of surrounding tumourous and non-tumourous cells owing to their growth and therefore patients with cancer show higher cfDNA concentrations in their blood. ¹⁴ cfDNA in healthy individuals mainly derives from haematopoietic cells through apoptosis. ¹⁵ cfDNA concentrations in plasma may be influenced by tumour stage, size and location. ¹⁶

Other non-neoplastic diseases resulting in increased cfDNA concentrations include myocardial infarction, severe viral infections, stroke, sepsis and severe trauma.^17–24 Furthermore, somatic alterations may be detected on cfDNA and it represents a valuable source of DNA if the primary tumour cannot be sampled. Interestingly, a recurrence of the primary tumour or metastases may be detected via cfDNA.^25,26 As necrosis preferably takes place in malignant processes and results in larger cfDNA fragments, malignant processes are assumed to lead to longer fragments.^27,28 Therefore, elevated levels of longer fragments of cfDNA may also serve as a good marker for the presence of DNA from malignant neoplasia. ²⁹ Nevertheless, the latest studies indicate that tumour-derived cfDNA is shorter than cfDNA derived from healthy cells.^30,31

Prior studies have shown that cfDNA concentration and integrity (ratio between long and short fragments) may harbour potential as a diagnostic and prognostic marker for primary and metastatic breast cancer and various other human cancers.^12,32 Interestingly, prior studies have shown a predictive capability of cfDNA concentration, integrity and/or specific alterations (mutations, promoter hypermethylations) in patients with cutaneous and ocular melanomas.^33–35 Burnett et al showed that a higher concentration of cfDNA in the plasma of dogs with lymphoma was associated with severity and prognosis of the disease. ³⁶ To our knowledge, cfDNA has never been studied in cats with (ocular) melanomas. Therefore, the aim of this study was to evaluate cfDNA concentration levels and fragmentation (integrity) in the blood of cats with FDIMs and cats with iris naevi, as well as cats without ocular abnormalities, in order to evaluate the use of cfDNA as a diagnostic and/or prognostic marker for cats with FDIMs.

Material and methods

Animals

This study was approved by the Austrian Federal Ministry of Science, Research and Economics, and owner consent obtained. Cats with and without melanocytic abnormalities of the iris, which were examined or treated at the Veterinary University of Vienna between 1 January 2015 and 10 August 2017, were enrolled in this study. Inclusion criteria for cats included a diagnosis of an iris melanoma or iris naevus, based on ophthalmological examination. Inclusion criteria for the control group included freedom from any ocular melanocytic lesions or other ocular tumours. The respective age, breed, sex, possible concurrent diseases and suspected metastases were recorded for all cats used in this study.

Samples

EDTA blood samples (2 ml) were obtained from cats with iris melanomas (n = 34), iris naevi (n = 30) or cats without melanocytic abnormalities of the iris (n = 32), and centrifuged at 1000 g for 10 mins at 4°C to separate blood cells from plasma. Plasma was carefully transferred to a 1.5 ml Eppendorf tube and centrifuged at 1000 g for 10 mins at 4°C. Plasma was carefully transferred to a new 1.5 ml tube without touching any cell pellets. Plasma samples were stored at −80°C until cfDNA extraction.

Ophthalmological examination

Both eyes were examined using slit-lamp biomicroscopy (Kowa portable slit-lamp SL-14; Kowa Optics) and direct and indirect ophthalmoscopy (Heine Omega 2C; Heine) by a board-certified ophthalmologist (BN). Intraocular pressure was measured using rebound tonometry (TonoVet, TV01; Dioptrix). Fluorescein dye and the Schirmer tear test were applied as necessary. Iris melanomas or naevi were characterised using the following criteria: colour and intensity, shape and extent of the melanocytic abnormality, iris surface and thickness in the respective area. The following features were considered indicative of melanoma: dark brown colour, irregular shape, invasion of other tissues than the iris (eg, ciliary body, sclera, lens, cornea), loss of normal iridial surface and thickening of the iridial tissue in the respective region.

Histopathological examination

Twenty-four eyes with clinical signs suggestive of iris melanoma were enucleated under anaesthesia. Within 1 h of enucleation the eyes were fixed in 4% formaldehyde for 24–72 h. Following embedding of the fixed eyes in paraffin, 2 µm sections were prepared. All sections were examined by ocular histopathologists following haematoxylin and eosin staining. If high melanin concentration of the tumour prevented an accurate diagnosis, bleached sections were prepared. Tumour spreading was graded using the grading system by Kalishman et al. ³⁷

cfDNA extraction, fragment length and concentration analysis

Plasma samples were thawed slowly at 4°C. Aliquots of 450 µl of plasma were diluted to 1 ml with 550 µl PBS (Bio Whittaker; Lonza) and cfDNA was immediately extracted using the QIAmp Circulating Nucleic Acid Kit (Qiagen) according to manufacturer’s instructions.

Primer and probes for feline amyloid beta precursor protein (APP) and beta actin (ACTB) were designed for amplification of amplicons of increasing length, using Primer Express 2.0 (Life Technologies). Each assay for APP and ACTB was designed with the same forward primer and probe but altered reverse primer. Specificity of all primer pairs was determined using the Basic Local Alignment Searching Tool of the NCBI database (http://www.ncbi.nlm.nih.gov/tools/primer-blast/index.cgi?LINK_LOC=BlastHome). Amplicon folding for each primer pair was reviewed with the Mfold Web Server (http://unafold.rna.albany.edu/?q=mfold/DNA-Folding-Form). ³⁸

All qPCR assays for fragment length and concentration determination were validated by a generation of standard curves and calculation of PCR reaction efficiencies using ViiA7 Software v1.1 (Life Technologies). All primer and amplicon details are listed in Table 1.

Table 1.

PCR primer for cell-free DNA and integrity analysis

Gene symbol	Amplicon number	Gene name	NCBI accession number	Oligo sequence (5′–3′)			Amplicon length (bp)	PCR reaction efficiency (%)	R²
				Forward primer	Probe	Reverse primer
APP	1	Amyloid beta precursor protein	NC_018731.1	AGACTCGCTTGGTCCTCAAAATA	FAM-CCGTGGACAGGCAGGGCGGATAC-BHQ1	TCAGGTTCACGCAGAACAAAA	75	92.40	0.998
	2					GAAAGCTGATTGGGCATGG	173	99.04	0.998
	3					ACTTAGGATGCGGCAATTAGCT	316	95.81	0.997
	4					CAGGGGATGCGGATAGTGC	459	92.97	0.994
ACTB	1	Actin beta	NC_018738.2	GGACCTGACGGACTACCTCATG	FAM-AGATCCTCACGGAGCGCGGCT-BHQ1	GCGGTGGTGGTGAAGCTG	63	92.77	0.992
						AGGGAGGACGAGGACGC	162	93.01	0.994
	2					AACCGAAGGCAGACTTAGCTTC	334	91.82	0.993
	3					GGCGGTGATCTCCTTCTGC	498	95.40	0.996
	4

bp = base pairs; FAM = 6-carboxyfluorescein; BHQ1 = black hole quencher 1

Real-time PCRs were carried out in 20 µl reactions containing 0.2 mM of each dNTP, 4.4 mM MgCl₂, 1× Buffer B2 (Solis BioDyne), 200 nM of each primer, 200 nM of probe, 50 nM ROX (Biotium), 1 unit HOT FIREPol DNA polymerase (Solis BioDyne) and 2 µl of each sample. All samples were analysed in duplicate using a ViiA7 Real-Time PCR System (Life Technologies) using the following temperature profile for assays with amplicon lengths <300 base pairs (bp; APP1, APP2, ACTB1, ACTB2): initial denaturation at 95°C for 10 mins, 50 cycles at 95°C for 15 s and 60°C for 1 min. For assays with amplicon lengths >300 bp (APP3, APP4, ACTB3, ACTB4) a step at 72°C for 1 min was added at the end of every cycle as shown in prior studies. Data were collected at 60°C or at the 72°C step for longer amplicons.

Amplicon concentrations were measured using ViiA7 Software v1.1 (Life Technologies) and subsequently converted into ng/ml plasma. The cfDNA concentration (APP1, ACTB1) and integrity index were calculated for both measured genes (APP/ACTB2–4 : APP/ACTB1).

Statistics

Statistical analysis was performed using IBM SPSS Statistics 20.0. Differences in the cfDNA concentrations and integrity levels between the three groups (iris melanoma, iris naevi and controls) were analysed using the Mann–Whitney U-test. Correlations between cfDNA concentrations or integrity levels and the age of the cats were tested using Spearman rho test. Furthermore, cfDNA concentration and integrity levels were compared to the different breeds, sex or concurrent diseases, using the Mann–Whitney U-test.

For all statistical tests a P value <0.01 was assumed to be significant.

Results

Two outliers were excluded from statistical analysis owing to unusually high levels of cfDNA; a potential cellular contamination could not be excluded in these cases. Therefore, 33 cats with iris melanoma, 30 cats with iris naevi and 31 control cats were included.

cfDNA concentrations between the three different groups (iris melanoma, iris naevi, controls) were not significantly different (P >0.01). Median concentrations and ranges (ng/ml) of their respective amplicons of APP and ACTB and integrity indices within the groups are shown in Table 2 and Figure 1.

Table 2.

Median results and ranges (ng/ml) of quantitative PCR of four amplicons for beta-actin (ACTB1–4) and amyloid-beta precursor protein (APP1–4) and their respective integrity levels (amplicon_x : amplicon₁)

Group		ACTB1	ACTB2	ACTB3	ACTB4	APP1	APP2	APP3	APP4	ACTB4/1	ACTB3/1	ACTB2/1	APP4/1	APP3/1	APP2/1
Melanoma	n	33	33	33	33	33	33	33	33	33	33	33	32	32	32
	Minimum	9.638	1.227	2.741	1.056	0.041	0.403	0.187	0.142	0.051	0.165	0.052	0.052	0.074	0.126
	Maximum	118.921	52.003	75.496	35.230	19.905	19.127	8.154	8.654	0.733	1.196	1.466	0.863	0.932	2.119
	Median	19.025	12.834	10.014	5.420	5.473	3.044	1.379	1.282	0.190	0.397	0.525	0.255	0.260	0.490
Naevus	n	30	30	30	30	30	30	30	30	30	30	30	30	30	30
	Minimum	4.987	0.707	1.505	1.400	0.916	0.228	0.215	0.211	0.033	0.109	0.113	0.055	0.043	0.102
	Maximum	155.380	32.874	25.506	14.989	45.052	10.800	6.714	6.091	0.711	1.396	1.221	0.446	0.897	1.365
	Median	24.282	14.013	10.763	4.097	4.291	2.101	1.017	0.753	0.150	0.411	0.521	0.178	0.217	0.462
Control group	n	31	31	31	31	31	31	31	31	31	31	31	31	31	31
	Minimum	11.039	6.000	2.002	0.699	1.371	0.858	0.218	0.193	0.029	0.054	0.349	0.049	0.055	0.251
	Maximum	82.997	50.099	24.493	27.025	31.318	10.354	9.814	6.912	0.766	0.706	1.297	0.588	0.395	1.310
	Median	31.969	20.663	11.780	5.489	6.734	3.273	1.467	0.993	0.156	0.384	0.659	0.183	0.214	0.505

Figure 1.

Cell-free DNA (cfDNA) (a,b) concentration and (c,d) integrity of cats with iris melanoma (n = 33), iris naevus (n = 30) and cats without ocular abnormalities (n = 31; ‘control’). Results were achieved using qPCRs for (a,c) ACTB and (b,d) APP with increasing amplicon lengths (ACTB1–4, APP1–4); cfDNA integrity was determined as the ratio of longer fragments (amplicon 2, 3 or 4) to the shortest fragment (amplicon 1). Statistical analysis of differences between the three groups was performed using the Mann–Whitney U-test and revealed P values >0.01. Extreme values are represented with an asterisk and outliers as an open circle

The overall median age of the cats of this study was 7 years (range 1–18 years). The age of two control cats was not recorded. Median ages of cats with iris melanomas, naevi or controls were 10 years (range 2–18 years), 8 years (range 2–16 years) and 5 years (range 1–13 years), respectively. No correlation between cfDNA concentrations or integrity index and age was found in this study (P >0.01).

The following breeds were included in this study: domestic shorthair (DSH; n = 73 [78%]), British Shorthair (BSH; n = 7; 7.5%), Maine Coon (n = 7; 7.5%), domestic longhair (DLH; n = 2 [2%]), Chartreux (n = 2; 2%), Persian (n = 2; 2%) and other breeds (n = 1; 1%). Cats with iris melanomas were either DSH (26/34; 79%), BSH (3/33; 9%), Maine Coon (3/33; 6%), DLH (1/33; 3%) and Persian (1/33; 3%). Of the 30 cats with iris naevi, 24 were DSH (80%), three were BSH (10%), one was a Chartreux (3%), one was a Persian (3%) and one a DLH (3%). Cats in the control group were DSH (24/31; 77%), Maine Coon (4/31; 13%), BSH (1/31; 3%), Chartreux (1/31; 3%) or other breeds (1/31; 3%). There were no significant differences in cfDNA concentrations between different breeds (P >0.01).

Thirty-two cats in this study were female spayed (34%), 14 were intact females (15%), 39 were neutered males (41%) and nine were intact males (10%). Among cats with iris melanoma there were 13 spayed females (40%), four intact females (12%), 14 neutered males (42%) and two intact males (6%). Cats with iris naevi included 10 spayed females (33%), three intact females (10%), 15 neutered males (50%) and two intact males (7%). The control group consisted of nine spayed females (29%), eight intact females (26%), nine neutered males (29%) and five intact males (16%). Intact males showed a significantly higher concentration of all ACTB fragments (ACTB1–4: 0.005, 0.01, 0.001, 0.007) and the second (APP2: 0.002) and third (APP3: 0.005) fragment of APP compared with neutered males; such an observation was not made in female cats (P >0.05). There was no significant difference in sex distribution between the three study groups (P = 0.23)

Twelve cats with iris melanoma, seven with iris naevi and three control cats had concurrent diseases, including heart or kidney disease or non-ocular tumours. Statistical analysis revealed no significant differences in cfDNA concentrations or integrity index associated with concurrent diseases (P >0.01).

Twenty-five of the 34 cats with suspected iris melanomas were enucleated. Histopathology revealed melanoma of the anterior uvea in 24 of the cases; one case was a melanoma affecting the posterior uvea. Kalishman grades for the iris melanomas included two grade I, 10 grade II, four grade III, two grade IV, six grade V and none with grade VI. There was no significant association between Kalishman grades and cfDNA concentration or integrity index.

Four cats with melanoma of the anterior uvea and the one cat with a melanoma affecting the posterior uvea had suspected (diagnostic imaging) or proven metastases (necropsy). No association between metastases and cfDNA concentrations or integrity levels of cats with feline intraocular melanomas (P >0.01) was identified in this study. All results of this study are shown in Supplementary Table 1 (supplementary material).

Discussion

Feline iris melanomas are the most common intraocular tumours in cats and show metastatic rates between 19% and 63%.^1,6,37 A confirmed diagnosis of this kind of tumour can only be achieved via histopathological examination of the enucleated eye or a biopsy of the respective lesion. Nevertheless, biopsies bear a significant risk for postoperative uveitis and may result in enucleation. Furthermore, preserving an affected eye owing to the lack of a diagnosis or a false diagnosis may result in a higher likelihood of distant metastasis. In cats with suspected iris naevi or other benign melanocytic iridal abnormalities, close monitoring for progression is advised in order to identify a neoplastic process early on.

Neoplastic cells release their DNA into the bloodstream of the patient owing to necrosis and apoptosis either by macrophages or directly.^12,29 These DNA molecules in plasma show a predictable fragmentation pattern, which has been related to cancer progression. ²² Increased release of cfDNA from tumour cells and adjacent non-tumour cells result in higher total cfDNA concentrations in human patients with various cancers than in healthy control individuals.^17,39–45 Furthermore, the integrity index (ratio between long and short fragments) harbours the potential as a biomarker for diagnosis and treatment surveillance in cancer patients.^41,43 Therefore, the aim of this study was to determine whether cats with iris melanomas show higher cfDNA concentration and/or a different fragmentation pattern, compared with cats with iris naevi, and without ocular abnormalities.

Healthy human individuals have a cfDNA concentration of between 0 and 100 ng/ml of blood (average 30 ng/ml) compared with 0–1000 ng/ml of blood (average 180 ng/ml) in patients with cancer.^12,46 Median results of cfDNA concentration in cats without ocular abnormalities, cats with iris naevi or iris melanomas were not significantly different. Nevertheless, cats with melanocytic abnormalities of the iris showed a higher range of cfDNA concentration with lower median concentration of cfDNA. However, a significant difference of cfDNA between these three groups was not found, although human cancer patients show highly significant differences in cfDNA levels compared with benign and control groups. ⁴⁷ Furthermore, dogs with lymphoma show a higher cfDNA concentration than clinically normal dogs. ³⁶ Interestingly, dogs with the highest levels of cfDNA among lymphoma patients had shorter remission times and showed a correlation between their cfDNA and clinical severity and prognosis. ³⁶ In this study, cats without ocular abnormalities showed partially higher concentrations of cfDNA than those with iris melanomas or naevi. Nevertheless, these results were not significant

cfDNA is present in different fragment sizes in the blood of individuals. The integrity index, the ratio of longer to shorter fragments, of cfDNA has been shown to be higher in metastatic breast cancer than in locally confined and benign lesions. ⁴⁸ Over 50% of breast cancer patients have short and long fragments in their plasma. ⁴³ Moreover, a higher integrity index in breast cancer patients and colorectal cancer patients is associated with several prognostic factors, such as lymphovascular invasion, lymph node metastasis and tumour size.^48,49 Prior publications have demonstrated an absence of cfDNA fragments >200 bp in healthy individuals or patients with benign tumours. ⁴⁷ Furthermore, an expression of 800 bp bands was associated with vascular invasion in cancer patients. ⁴⁷

Nevertheless, no correlation was found between clinicopathological parameters (tumour size, stage, grade, metastases) and cfDNA concentration or integrity. Feline iris melanomas, which were sent for histopathological examination, were graded based on spreading and prognosis according to Kalishman et al. ³⁷ Kalishman grading was not associated with cfDNA concentrations or integrity index. Furthermore, cats with suspected or proven metastases through either diagnostic imaging or necropsy did not show higher cfDNA levels than those without. Therefore, cfDNA, as shown by Zaher et al, ⁴⁷ is an insufficient clinicopathological and also histopathological marker for feline uveal melanomas.

Whereas no association between breed, age or concurrent diseases and cfDNA was found in this study, intact males showed higher cfDNA concentrations than neutered males. A possible explanation for this finding may be male hormones resulting in higher apoptotic rates and causing higher cfDNA rates. To our knowledge no studies on changes in cfDNA concentrations related to male hormones are available.

There were some limitations to this study: as cats are significantly smaller than humans and most dogs, the plasma volume used for cfDNA extraction in this study was relatively low (450 µl). The resulting cfDNA amounts extracted were therefore less than in human patients. Furthermore, blood collection and storage of blood or plasma may have also influenced the results through possible fragmenting of the cfDNA or contamination with DNA of white blood cells, as shown in prior studies. ⁵⁰ Nevertheless, all blood samples were centrifuged within 2 h of drawing blood, and plasma samples were stored at −80°C to minimise cfDNA fragmentation and contamination during the study.

Moreover, Bellhorn showed that molecules with an effective diffusion radius of up to 8.5 nm are able to pass through the blood aqueous barrier (BAB) in feline eyes; ⁵¹ concurrent inflammation may open the BAB even more. To our knowledge, there are no studies on the ability of cfDNA to pass the BAB; only one prior study has shown the presence of tumourous cfDNA in humans with metastasising uveal melanomas. ³⁵ In the aforementioned study, cfDNA may have been derived from metastases rather than the primary tumour. Considering that only cats with more progressive FDIMs had concurrent inflammation, we conclude that a lack of breakdown of the BAB may prevent higher cfDNA levels, especially of longer fragments in the plasma of cats with FDIMs.

Our results suggest that the levels of cfDNA potentially released from FDIM cells may be too low for detectable effects on total cfDNA amounts in the blood circulation of the patients. Furthermore, the lack of a statistical significant difference between cfDNA of cats with iris naevi and melanomas may be owing to the fact that the diagnosis of some FDIMs and all iris naevi was based solely on clinical examination, which may have been subject to misinterpretation. For future studies, aqueous humour may be investigated for the presence of cfDNA owing to the close contact with the primary tumour. Extraction kits for small sample amounts, such as aqueous humour, are currently not available; however, this may become available in the future. In addition, the detection of circulating melanoma cells in humans with metastasising cutaneous melanomas also showed promising results in diagnosis, monitoring and managing of this disease. ⁵² Therefore, the detection of circulating melanoma cells in plasma or aqueous fluid of (metastasising) FDIMs may be another new diagnostic possibility to be investigated in the future.

Conclusions

Overall, cfDNA concentration and integrity levels seem to be insufficient as diagnostic or prognostic markers for cats with feline iris melanomas. Although humans with metastatic diseases usually show higher cfDNA concentrations and integrity levels than healthy individuals or patients with benign tumours, cats with metastases had no significant differences in their cfDNA levels compared with those without metastases. We conclude that cfDNA is inadequate as a marker for iris melanomas in cats.