Serum DJ-1/PARK 7 Is a Potential Biomarker of Choroidal Nevi Transformation

Abstract

Purpose.: There is substantial evidence that intraocular melanomas arise from benign nevi in the uveal tract. Previous studies performed by the authors revealed that uveal melanoma cells secrete the oncoprotein DJ-1/PARK7 into the extracellular environment and circulation. The aim of this study was to determine whether circulating DJ-1 serum levels correlate with known clinical risk factors of nevi growth.

Methods.: Standardized ultrasonography, optical coherence tomography, and eye fundus examinations were used to evaluate the clinical risk factors of nevi growth. These clinical risk factors (including nevi size, distance of margins to the optic disc, detection of acoustic hollowness, presence of ocular symptoms, orange pigment, subretinal fluid, and absence of drusen) were examined in 53 consecutive patients from January 2009 to February 2011. Serum levels of DJ-1/PARK7 in these patients and in healthy age- and sex-matched controls (n = 32) were analyzed using ELISA.

Results.: Within the choroidal nevi group, DJ-1 serum levels were higher in those with symptoms (P < 0.033), with a nevus thickness greater than 1.5 mm (P < 0.001), a large basal diameter greater than 8 mm (P < 0.001), and the presence of acoustic hollowness (P < 0.001), compared to those patients without these risk factors. Similar significant differences were found when these at risk nevi subgroups were compared to healthy persons.

Conclusions.: Elevated serum levels of DJ-1 are associated with choroidal nevi transformation risk factors. Therefore, DJ-1 appears to be a promising factor for predicting the growth of choroidal nevi and may be a potential biomarker of malignancy.

Malignant uveal melanoma (UM) is the most common primary intraocular tumor in adults. UM occurs in approximately 4 to 7 persons per million in the United States and is most prevalent in the Caucasian population. ¹ Unlike cutaneous melanoma, UM disseminates primarily through the bloodstream and preferentially metastasizes in the liver. Even with the successful treatment of UM primary tumors, patients are still at risk for metastases more than 20 years after the initial diagnosis. ² In the Collaborative Ocular Melanoma Study (COMS), the Kaplan-Meier estimates of 2-, 5-, and 10-year metastasis rates were 10%, 25%, and 34%, respectively; however, only 0.24% (21/8712) of the patients exhibited detectable metastases at the time of diagnosis. ³ Despite new advances in controlling liver metastatic disease, metastatic lesions are generally difficult to manage, especially in those cases with multiple nodules. Nevertheless, evidence shows that hepatectomy in patients with a solitary metastasis extends life expectancy. ⁴  

Initially, choroidal nevi, also referred to as pigmented lesions or small choroidal melanocytic tumors, were routinely considered to be benign tumors; however, they may transform into malignant melanomas and eventually develop into metastatic disease. ⁵ The prevalence of nevi in the Caucasian population is similar among men and woman (8.9% and 8.3%, respectively), and the risk of malignant transformation is estimated to occur in 1 in 8845 persons. ⁶ Independently of its incidence, the growth of nevi is considered a strong indicator of malignancy in the ocular oncology literature. ^7,8 It is believed that malignant choroidal melanomas originate in a preexisting benign nevus and that these tumors undergo a gradual change from a differentiated to a less differentiated state. ⁹ In this context, the main challenge for ophthalmologists is to identify the clinical features associated with the growing lesions. The identification of these risk factors improves the ophthalmologist's ability to estimate the risk of future growth and malignancy. ^{8,10 –12} Predictive factors for developing melanoma include the presence of the following ocular symptoms: tumor size, visible orange pigment under the overlaying retina, existence of subretinal fluid, absence of drusen, tumor margin within 3 mm of the optic disc, and detection of acoustic hollowness by ultrasonographic examination. ^8,12 Thus, clinical differences between UM and nevus, based only on size criteria, may not be accurate. Better understanding of the specific clinical and molecular factors for predicting nevus transformation would improve life expectancy during the early conservative treatment of patients. 

DJ-1 was discovered 14 years ago as an oncogene with transforming properties in cooperation with H-ras. ¹³ It is considered a multifunctional oxidative stress response protein that defends cells against reactive oxygen species and mitochondrial damage, even though the details of its biochemical function remain unclear. ¹⁴ During the past decade, DJ-1 has been implicated in several biological processes, such as fertilization, ¹⁵ mRNA stabilization in response to cAMP, ¹⁶ androgen receptor signaling regulation, ^17,18 Parkinson's disease, ¹⁹ and cancer. ^{20 –23} We previously showed that DJ-1 was overexpressed and secreted by human UM cells in culture and, more relevantly, identified circulating DJ-1 in the sera of UM patients, thus suggesting its potential use as a biomarker for choroidal nevi transformation. ²⁴  

We investigated the possible correlation between predictive factors of nevi malignant transformation (presence of symptoms, size, orange pigment, existence of subretinal fluid, absence of drusen, tumor margin within 3 mm of the optic disc, and acoustic hollowness) and patient serum levels of DJ-1. 

Patients and Methods

Patients and Clinical Procedures

Fifty-three consecutive patients with diagnoses of choroidal nevi were followed up between January 2009 and February 2011 in the Ocular Oncology Unit at the Complexo Hospitalario Universitario de Santiago. The clinical diagnoses of these patients were made based on a complete ophthalmic examination and standardized ocular ultrasonography. All patients were under observation and had not previously received any treatment. The inclusion criterion for the patient group was defined by the presence of one nevus up to 3 mm thick and a large basal diameter (LBD) <12 mm, as estimated by ocular ultrasonography or by fundus photographs. Thickness measurement was determined by means of ultrasonography, with the exception of 17 patients whose thickness was not high enough to be detected by the ultrasound signal. In these latter cases, it was assumed that their thickness was not 0.5 mm. Exclusion criteria included antecedents of any type of cancer, presence of suspicious cutaneous melanocytic lesions, central nervous system illness, and stroke. Moreover, 32 healthy volunteers obtained from routine ophthalmologic examinations were selected as a control group. This group, matched in age and sex distribution with the patient group, was subject to the same exclusion criteria. Participants from both groups were exposed to identical clinical procedures. Both were included in the study after giving informed consent according to the Declaration of Helsinki. This study was also approved by the Comité Ético de Investigación Clínica de Galicia. 

Standardized ultrasonography was carried out in the patients (I3-ABD System; Innovative Imaging Inc., Sacramento, CA) by means of a 10-MHz probe. After instillation of an anesthetic drop, topographic, quantitative, and kinetic ultrasound was performed directly over the eye using a coupling gel (Viscotears, carbomer gel 2%; Novartis, Barcelona, Spain). The same physician carried out this examination in all patients, and the following acoustic parameters were studied: tumor location, LBD and thickness measurement, and acoustic hollowness. The existence of subretinal fluid related to the tumor was examined (Cirrus SD-OCT; Carl Zeiss Meditec, Inc., Oberkochen, Germany) when possible. If OCT was not applicable, the presence of subretinal fluid was evaluated by ophthalmoscopy and ocular ultrasonography. 

Other related clinical signs of prognostic value for nevi transformation were evaluated in all cases. These included the presence of ophthalmic symptoms directly related to the tumor (floaters, photopsias, and visual acuity diminution), drusen affecting the overlying retina, presence of orange pigment, and a tumor margin within 3 mm of the optic disc (Supplementary Table S1). 

Blood samples for DJ-1 serum detection were collected in a tube (STT II Advance Vacutainer; Becton Dickinson, Franklin Lakes, NJ), allowed to clot for 30 minutes, and centrifuged (15 minutes/1500g) to separate the serum. Immediately, serum samples were aliquotted, properly coded, and frozen at −80°C until processing. Double masking was performed between the clinical information and the data collected from the patients' serum. 

Enzyme-Linked Immunosorbent Assay

The quantitative measurement of human DJ-1/PARK-7 in human serum samples was analyzed using an ELISA kit (CircuLex Human DJ-1/PARK-7 ELISA Kit; MBL International Corporation, Woburn, MA) according to the manufacturer's instructions. Absorbance from each sample was measured in duplicate using a spectrophotometric microplate reader, at dual wavelengths of 450/540 nm (Versamax Microplate Reader; Associates of Cape Cod Incorporated, East Falmouth, MA). 

Statistical Analysis

DJ-1 serum concentration levels are expressed as the mean ± SD. Differences in DJ-1 values between the patient group lacking risk factors, the patient group with risk factors, and controls were statistically analyzed using the Student's t-test for unpaired data. When the number of samples was small or the distribution was not normal, the nonparametric Mann-Whitney U test was applied. The Benjamini-Hochberg method was also used to adjust for multiple comparisons. ²⁵ In addition, receiver operating characteristic (ROC) analysis ²⁶ was performed, and the corresponding area under the curve (AUC) was calculated to evaluate the value of DJ-1 as a biomarker for the presence of risk factors. To carry out the statistical analysis, a subdivision in size, thickness, and major LBD was performed. In relation to thickness, we considered major nevus those >1.5 mm and minor nevus those <1.5 mm. For LBD, we subdivided this group into major nevi (8 mm) and minor nevi (<8 mm). The relationships among LBD, thickness measurements, and DJ-1 levels were also analyzed as continuous variables. To avoid the need for imposing arbitrary parametric effects, generalized additive regression models (GAMs) were applied. ²⁷ Finally, multivariate regression with a stepwise procedure based on the Akaike Information Criterion ²⁸ was used to select the best subset of dichotomous explanatory risk factors. All these analyses were performed using the R software, version 2.12.0. ²⁹ GAMs were fitted using the mgcv package, ³⁰ and the ROC analysis was performed using the pROC package. ³¹ In all cases, P < 0.05 was considered statistically significant. 

Results

Clinical Findings

Fifty-three patients (25 men, 28 women) with diagnoses of nevi were recruited for this study. Their average age was 65.54 years and ranged from 23 to 91. The control group was made up of 17 men and 15 women. Their average age was 66.63 years and range between 26 and 98 years. No statistically significant differences were associated with age or sex distribution between both groups. Areas in which nevi were found were as follows: 30.18% in the equator, 28.30% on the posterior pole, 24.54% next to the optic nerve (<3 mm), and 16.98% on the periphery. The average LBD was 5.61 mm (±2.81 SD). In the group of measurable nevi (n = 38), the average thickness was 1.30 mm (±0.65 SD). 

DJ-1 Serum Levels in the Patient Group

When the average serum values of DJ-1 between patients (37.39 ng/mL, ±24.99 SD) and controls (32.98 ng/mL, ±16.30 SD) were compared, no statistically significant differences were found (Student's t-test, P = 0.892) (Table 1). Nevertheless, an abnormal distribution with a large right tail was observed in the nevi patients compared with the control group (Fig. 1). However, variations in DJ-1 levels were significant within the patient group. Interestingly, DJ-1 appeared to be more elevated in the sera of patients with symptoms at the time of diagnosis. These patients had an average concentration of 69.79 ng/mL (±28.62 SD) compared with asymptomatic patients, who had an average concentration of 34.75 ng/mL (±23.04 SD; Mann-Whitney U test, P = 0.033; AUC = 0.852). In addition, we found that DJ-1 serum levels were significantly higher in patients with nevi thickness >1.5 mm (73.22 ng/mL; ±28.36 SD) compared with those with nevi thickness <1.5 mm (29.06 ng/mL; ±14.97 SD; Mann-Whitney U test, P < 0.001; AUC = 0.916). Similarly, LBD with diameters >8 mm exhibited increased values of DJ-1 in relation to those with <8 mm (>8 mm, 73.03 ng/mL, ±24.55 SD vs. <8 mm, 29.11 ng/mL, ±16.48 SD; Mann-Whitney U test, P < 0.001; AUC = 0.928). Moreover, the relationship between the existence of acoustic hollowness and circulating DJ-1 levels was also statistically significant. Those patients with hollowness exhibited the highest average DJ-1 levels (86.95 ng/mL) compared with those without this risk factor (86.95 ng/mL, ±16.35 SD vs. 28.58 ng/mL, ±13.03 SD; Mann-Whitney U test, P < 0.001; AUC = 0.997). However, no statistical significance was found when we compared DJ-1 levels in the nevi group with the rest of the risk factors (Table 1). 

Table 1.

 

DJ-1 Levels in Patients and Controls and ROC Curve Analysis

Risk Factors	DJ-1 (ng/mL)
	Sample Size	Mean	SD	95% CI	P	AUC
Symptomatic
No	49	34.75	23.04	28.13–41.37	0.033 0.029	0.852 0.888
Yes	4	69.79	28.62	24.24–115.34
Control	32	32.98	16.30	27.10–38.86
Orange pigment
No	43	36.87	23.99	29.48–44.25	1 0.894	0.500 0.516
Yes	10	39.66	30.28	18.00–61.32
Control	32	32.98	16.30	27.10–38.86
Drusen
No	20	36.85	20.43	27.29–46.41	0.648 0.824	0.567 0.517
Yes	33	37.72	27.69	27.90–47.54
Control	32	32.98	16.30	27.10–38.86
Acoustic hollowness
No	45	28.58	13.03	24.67–32.50	<0.001 <0.001	0.997 0.984
Yes	8	86.95	16.35	73.28–100.62
Control	32	32.98	16.30	27.10–38.86
Optic disc
No	39	39.82	27.99	30.74–48.89	0.848 0.633	0.531 0.485
Yes	14	30.64	11.93	23.75–37.53
Control	32	32.98	16.30	27.10–38.86
Thickness (>1.5 mm)
No	43	29.06	14.97	24.45–33.67	<0.001 <0.001	0.916 0.872
Yes	10	73.22	28.36	52.93–93.51
Control	32	32.98	16.30	27.10–38.86
LBD (>8 mm)
No	43	29.11	16.48	24.03–34.18	<0.001 <0.001	0.928 0.894
Yes	10	73.03	24.55	55.46–90.59
Control	32	32.98	16.30	27.10–38.86
Subretinal fluid
No	45	36.80	24.99	29.29–44.30	0.648 0.592	0.580 0.605
Yes	8	40.74	26.46	18.63–62.86
Control	32	32.98	16.30	27.10–38.86

 Patients were grouped according to risk factors. 95% CI, 95% confidence interval.

Figure 1.

 

Serum DJ-1/PARK7 levels distribution within the (left) choroidal nevi and (right) control groups.

Results from the multivariate analysis showed that the best subset of dichotomous predictive risk factors explaining the levels of DJ-1 included the presence of symptoms, nevi thickness greater than 1.5 mm, and the presence of acoustic hollowness (R ² = 0.76) (Table 2), whereas the presence of acoustic hollowness was the most strongly associated risk factor. It should be noted that values of LBD larger than 8 mm were highly associated with both thickness values greater than 1.5 mm (odds ratio [OR] = 65.92; Fisher-test, P < 0.001) and the presence of acoustic hollowness (OR = 79.29; Fisher-test, P < 0.001). This could explain why this risk factor is not statistically significant when included in the multivariate regression model. 

Table 2.

 

Results of Multivariate Regression Analysis in the Patient Group

Risk Factors	Coefficient	SE	P	R 2
Symptomatic	17.88	7.11	0.015
Acoustic hollowness	42.91	7.79	<0.001	0.76
Thickness (>1.5 mm)	14.58	6.85	0.038

Results of the GAM analysis, which considered nevi thickness and LBD without categorization, are presented in Figure 2. Levels of DJ-1 in the sera of patients tended to increase as the thickness of the nevi increased, but only until 2 mm. Interestingly, for nevi greater than 2 mm, the levels of DJ-1 seemed to decline progressively (P = 0.004, R ² = 0.459). With respect to LBD, a positive, almost linear relationship existed with the levels of serum DJ-1 (P < 0.001, R ² = 0.233). It should be noted that the small R ² obtained in both cases suggests that the DJ-1 levels cannot be completely explained by basal diameter or nevi thickness but may be better explained by association with some other risk factors, such as acoustic hollowness. 

Figure 2.

 

Nonparametric estimate of the relationship between DJ-1 and (left) thickness or (right) base (solid line), along with pointwise 95% confidence intervals (dotted line).

It is important to emphasize the transformation of three nevi to uveal melanoma during the 24 months of the study (increased >0.5 mm in 6 months) (Supplementary Table S2). These patients were treated with ¹²⁵I brachytherapy. The average baseline serum level of DJ-1 in these three patients was 89.56 ng/mL. The individual levels for these patients were 62.96, 104.37, and 101.34 ng/mL, which were all higher than the average of the nevus group (37.39 ng/mL) and the control group (32.98 ng/mL). Exclusion of these three patients from the analysis did not change the significance or the magnitude of the analyses performed here (Supplementary Table S3). 

Comparison of DJ-1 Serum Levels between Patients and Control Groups

We also compared DJ-1 serum levels between the subgroups of nevi patients with risk factors and the control group. In this case, statistically significant differences were found between the control group and the group of patients with the following risk factors: symptomatic (P = 0 0.029), nevi thickness greater than 1.5 mm (P < 0.001), LBD larger than 8 mm (P < 0.001), and existence of acoustic hollowness (P < 0.001) (Mann-Whitney U test for symptomatic). With ROC analysis, the corresponding AUCs, when compared with the control group, were 0.888 for symptomatic patients, 0.872 for patients with thickness 1.5 mm, 0.894 for patients with LBD larger than 8 mm, and 0.984 for patients with acoustic hollowness. Clinical variables, such as the presence of orange pigment, absence of drusen, presence of subretinal fluid, and tumor margin within 3 mm of the optic disc, did not correlate with circulating levels of DJ-1 (Table 1). 

Discussion

In the present work, we show for the first time a correlation between known clinical risk factors of nevi growth and circulating DJ-1 levels. Previously, we showed that UM cells secrete DJ-1 into the extracellular environment and into the bloodstream. ²⁴ To discern the potential role of DJ-1 as a nevi transformation biomarker, we assayed circulating levels of DJ-1 in patients whose nevi were under observation. The results of this study illustrate a significant correlation between DJ-1 serum levels and the presence of the following risk factors: nevus thickness greater than 1.5 mm, LBD larger than 8 mm, and presence of acoustic hollowness. In support of these correlations, multivariate analyses reinforced the strong, independent entailment between DJ-1 levels and these three clinical risk factors for malignant transformation. This is the first report linking circulating DJ-1 protein levels with nevi growth. We believe that measuring DJ-1 serum levels may be used as a potential predictive factor for malignancy. 

Ocular ultrasonography is an imaging technique considered essential for accurately diagnosing and following up on a variety of intraocular tumors. ^32,33 In recent years, two ultrasonographic observations, nevi thickness >2 mm and acoustic hollowness, were shown to be predictive factors for uveal nevi growth and indirect malignant transformation. ⁸ Obviously, nevi thickness is a clinical observation directly related to the growth of a tumor. ³⁴ On the other hand, acoustic hollowness is a characteristic ultrasonographic observation in UM. ³³ This finding is related to the resultant ultrasound beam attenuation of the homogeneous and dense histology of these types of tumors. This type of attenuation is seen in B-mode echograms as hypoechogenic areas that are normally located near the base of the tumor and in A-mode ultrasonography as a progressive diminution in the ultrasonographic signal (kappa angle). ³⁵ Both thickness estimation and occurrence of acoustic hollowness are typical clinical signs often studied in standard clinical protocols for studying uveal nevi and UM. ³⁶ Previous proteomics studies performed by our group identified, for the first time, the overexpression of the oncoprotein DJ-1/PARK7 in UM. ^24,37 Furthermore, we discovered that the tumor mass was capable of secreting the soluble form of this protein both in vitro and in vivo, suggesting its potential use as an UM biomarker. ^24,38 Accordingly, preliminary results from our studies analyzing tumor fragment secretomes revealed high concentrations of DJ-1. Paralleling this in vitro secretion, we also detected abundant levels of soluble DJ-1 in the same sera from these patients as well as strong immunoreactive staining of this protein in each tumor mass (unpublished results, 2011). Furthermore, increased levels of DJ-1 were observed in several other types of tumors ^39,40 ; however, its secretion into the circulation was described only in breast cancer patients. In that study, DJ-1 protein was detected at high levels in the sera of only 37% patients with new diagnoses. ⁴¹ Therefore, in the present work, we chose to exclude patients with any other disease type previously associated with DJ-1. It is well documented that DJ-1 is implicated in a variety of conditions, such as Parkinson's disease, infertility, ischemic injury, and amyotrophic lateral sclerosis. ^18,19,42 Although there is no information associating DJ-1 and cutaneous melanoma, we decided to exclude patients with any suspicious cutaneous melanocytic lesions. It is well known that some cases of uveal nevi are associated with dysplastic nevi of the skin. ⁴³  

Based on our previous observations and experience with UM, we hypothesized that early-stage UM and nevi at clinical risk for transformation might also secrete DJ-1 in sufficient amounts for detection using a simple blood test. The role of DJ-1 in tumorigenesis or metastasis is not yet well understood, and its detection at higher levels in patients compared with healthy controls might indicate the use of the protein as a biomarker for tumor progression. The fact that we did not find any statistical variation in DJ-1 sera levels when comparing age- and sex-matched controls with patients was initially surprising. However, when we compared the group of patients at higher risk, based on known predictive factors, with control subjects, we found statistically significant differences. We believe that the mixture of low-risk nevi patients with those at high risk within the same group may explain this occurrence. As happens with many other known disease biomarkers in routine clinical use, DJ-1 can also be detected in healthy populations at low levels regardless of age or sex. ⁴² In fact, we detected similar levels of circulating DJ-1 in the control group as previously described. ⁴⁴ Knowing that it has been described that DJ-1 protein is subject to multiple oxidative modifications under stress conditions, ¹⁴ it can be considered that most oxidized forms of the protein may be characteristic of nevi at risk of growth, thus becoming a better biomarker than total DJ-1. We have performed preliminary assays by two-dimensional electrophoresis Western blot analysis showing different DJ-1 spot patterns comparing the sera of healthy subjects and patients; this research is being extended with the participants presented in this study plus uveal melanoma patients and will be presented in a future paper. 

Although we observed that few nevi exhibit a positive linear relationship with their thickness, there was a significant relationship between DJ-1 levels and both thickness and LBD according to GAM analyses (Fig. 1). This result may suggest that the activity of DJ-1 secretion may be influenced by more than one biological variable. Our group is investigating the relationship between DJ-1 levels and UM progression and correlating these findings with systemic disease, cell type, or both. Interestingly, we observed in three of our patients that high levels of circulating DJ-1 predicted nevi growth before its detection with conventional diagnostic tests; however, this result should be considered cautiously given the small sample size. Further analyses using larger sample sizes and further follow-up are essential for determining the use of DJ-1 levels as a biomarker for nevi transformation. 

In summary, elevated serum levels of the oncoprotein DJ-1 correlate with risk factors associated with choroidal nevi, such as presence of symptoms, size, and acoustic hollowness. DJ-1 appears to be a promising predictive factor for choroidal nevi growth and a potential biomarker for malignancy. 

Supplementary Materials

Table st1, PDF - Table st1, PDF 

Table st2, PDF - Table st2, PDF 

Table st3, PDF - Table st3, PDF 

Footnotes

 Supported by Grant PS07/09 from Consellería de Sanidade, Xunta de Galicia, Spain. MP is a Miguel Servet Fellow (Instituto de Salud Carlos III/SERGAS).

Footnotes

 Disclosure: M.F. Bande, None; M. Santiago, None; M.J. Blanco, None; P. Mera, None; C. Capeans, None; M.X. Rodríguez-Alvarez, None; M. Pardo, None; A. Piñeiro, None

The authors thank Stephen Gee (Johns Hopkins University School of Medicine) for editing this manuscript.