HLA Class I and II Genotype in Uveal Melanoma: Relation to Occurrence and Prognosis

Abstract

purpose. To assess whether human leukocyte antigen class I and class II alleles confer susceptibility to uveal melanoma or are related to specific clinical or tumor characteristics and survival.

methods. Between 1990 and 2004, 235 consecutive Dutch patients with diagnoses of primary uveal melanoma were typed for HLA class I and II, either by complement-dependent cytotoxicity test or by DNA-based technique. Allele frequencies were compared with those of a control group that consisted of 2440 healthy Dutch blood donors. In addition, allele frequencies of 138 patients with uveal melanoma, who underwent enucleation as primary treatment, were compared for tumor characteristics and survival.

results. With regard to tumor characteristics, correlations between HLA-DR13 and tumor size, HLA-B35 and spindle cell type, and HLA-B60 and ciliary body involvement were observed before correction for the number of alleles tested. Correlation was found between the presence of HLA-B44 and decreased survival. We did not find any allele that correlated with susceptibility to uveal melanoma after correction for the number of comparisons between patients and controls.

conclusions. This study shows that HLA class I and II antigens do not contribute to an increased genetic susceptibility to uveal melanoma. This does not exclude an important role for HLA antigens in immune surveillance against uveal melanoma and their metastases.

The human leukocyte antigen (HLA) complex, which is located within the major histocompatibility complex on chromosome 6, is one of the most polymorphic regions of the human genome. It encodes the HLA alloantigens, which play a central role in immune recognition, both with regard to T lymphocytes (CTLs) and to natural killer (NK) cells. The function of these cells is directly influenced by expression of HLA class I and II antigens on the tumor cells. ^{1 2} CD4⁺ and CD8⁺ T cells can only be activated when they recognize tumor-specific antigens in association with the appropriate HLA molecules. In many types of tumors, downregulation of HLA expression occurs as an effective mechanism to escape immune surveillance. ^{1 3} NK cells, on the other hand, specifically kill cells lacking HLA class I antigens. The most successful tumor cells are those that can downregulate HLA in such a way that they are not recognized by CTLs but still have sufficient HLA expression to avoid killing by NK cells. In uveal melanoma, the situation is completely different from that of most other tumors because the downregulation of HLA class I and II on the primary tumor is not associated with tumor cell escape and disease progression; rather, it is associated with favorable patient outcome. ^{4 5} This suggests that NK cells play a protective role in the development of metastatic disease by killing tumor cells with low HLA class I expression. ⁶  

Correlations between the presence of certain HLA alleles and the occurrence of diseases have been described, as in autoimmune disorders. One of the strongest HLA- disease associations is that of the class I allele, HLA-B27, and ankylosing spondylitis, in which approximately 96% of affected persons have an HLA-B27 allele. ⁷ Furthermore, correlations between HLA antigens and cutaneous melanoma have been reported. HLA-B40, -DR4, and -DR5 were found to be related to the occurrence of cutaneous malignant melanoma, ^{8 9} whereas HLA-B40 and the class II alleles HLA-DR11 and -DQ7 were related to local recurrence. It was suggested that class II genes influence cytokine production and thus influence local immune response against metastases. ¹⁰  

Similarly, the relationship between certain HLA antigens and the occurrence of uveal melanoma has been studied. In 1977, Ignatov et al. ¹¹ reported decreased frequencies of HLA-A25 and -B16 and increased frequencies of HLA-A3 and -A29 in 47 patients with choroidal melanoma. ¹² Later, Bertrams et al. ¹³ and Martinetti et al. ¹⁴ reported an increased frequency of HLA-A32. Völker-Dieben et al. ¹² observed an absence of the HLA-B27 allele in pure spindle cell tumors. Dieckhues et al. ⁸ found elevated frequencies of HLA-A1 and -B27 in patients with uveal melanoma, and Jager et al. ¹⁵ observed an association between the presence of HLA-B40 and death attributed to the metastasis of uveal melanoma. 

Therefore, the aim of this study was to assess not only whether HLA class I and class II alleles are related to the occurrence of uveal melanoma but especially whether the presence of any specific antigen is related to specific clinical or tumor characteristics or survival. 

Methods

HLA-typed Patients and Controls

Between 1990 and 2004, 235 consecutive Dutch patients with diagnoses of primary uveal melanoma were typed for HLA class I and II in the Department of Ophthalmology at the Leiden University Medical Center, either by complement-dependent cytotoxicity test or by DNA-based technique. Of the 235 patients, 128 were women and 137 were men. Ages ranged from 28 to 95 years, with a mean age at time of diagnosis of 62 years. Of the eyes with melanoma, 111 underwent enucleation as primary therapy and 112 received ruthenium brachytherapy; of the latter, 70 tumors additionally received transpupillary thermotherapy and 10 received transscleral thermotherapy. Two cases were treated expectatively. Of the 184 eyes that were not treated with enucleation as primary therapy, 27 were enucleated at a later time because of tumor progression or a blind and painful eye. The control group consisted of 2440 healthy Dutch blood donors. For this group, HLA gene frequencies are extensively analyzed and well controlled, and this group is considered to be a good representation of the HLA antigen distribution in the Dutch population. ¹⁶  

HLA Typing

The control group was typed for HLA by serologic methods using the standard National Institutes of Health complement–dependent lymphocytotoxicity test for HLA-A and -B typing ¹⁷ and the propidium iodide method for HLA-DR, and -DQ typing. ¹⁸ HLA type was determined on peripheral blood leukocytes, either by serologic methods or by PCR-amplified DNA, using sequence-specific oligonucleotides. ¹⁹  

Tumor Specimens

Formalin-fixed, paraffin-embedded specimens from 138 patients with uveal melanoma were obtained and available for this study. The lesions came from 72 female and 66 male patients whose average age was 65 years (range, 28–95 years). Each tumor sample was processed for conventional histopathologic diagnosis. The research protocol followed the tenets of the Declaration of Helsinki. 

Histopathologic Examination

Histologic sections were prepared from tissues fixed in 10% buffered neutral formalin for 48 hours and were embedded in paraffin. Hematoxylin- and eosin-stained 4-μm sections were prepared through the central region of the tumor and were reviewed by an ocular pathologist for intraocular localization, cell type, largest basal diameter, and prominence. 

Statistical Analysis

Woolf-Haldane analysis ^{20 21} was used to assess the significance of differences in the HLA antigen frequencies in patients with uveal melanoma compared with the control population. Odds ratios were calculated as an estimation of the relative risks. P values were calculated using Fischer’s exact test. Correction of P values for the number of comparisons was performed using the method of Edwards. ²² Survival analyses for tumor characteristics were performed by using Kaplan-Meier analysis and log rank test. 

Results

Distribution of HLA Class I and II Alleles

In 235 Dutch patients with uveal melanoma and 2440 Dutch healthy blood donors, the frequencies of HLA class I (A and B) and class II (DR, DQ) alleles were analyzed. In total, 63 different alleles were determined. An increased frequency was found for the HLA-DQ4 allele (frequencies, 6% patient group, 3% control group; P = 0.033) before correction. After correction for the number of comparisons, the P value was close to 1.000 for this and all other HLA alleles. In Tables 1 and 2 , the distribution of the HLA antigens is summarized. 

Survival Analysis and Tumor Characteristics

To assess whether HLA class I and II alleles were related to specific clinical or tumor characteristics and survival, allele frequencies of 138 patients with uveal melanoma who underwent enucleation as primary treatment were compared regarding tumor characteristics and survival. 

Mean survival was 67.1 months (range, 3–354 months). Tumor size, location, and cell type of the tumor were analyzed for correlation with patient survival (Table 3) . Subsequently, comparisons were made between alleles that appear in the general population at a frequency of >10% and survival. Decreased survival was found for patients with the HLA-B44 allele (P = 0.012; Fig. 1 ); a similar trend could not be found for the other alleles tested. Tumor size was classified into three groups according to Collaborative Ocular Melanoma Study (COMS) criteria. ²³ Categories were small tumors (apical height, 1.50–2.4 mm; largest basal diameter [LBD], 5–16 mm), medium-sized tumors (apical height, 2.5–10.0 mm; LBD, ≤16 mm), and large tumors (apical height, >10 mm or LBD >16 mm). Thirty-one tumors were classified as small, 155 as medium, and 49 as large. Mean survival of patients with, respectively, a small, medium, and large tumor was 86.4 (SD, ±41.8) months, 70.9 (SD, ±48.0) months, and 43.5 (SD, ±34.7) months. Survival analysis showed that tumor size was significantly correlated with survival (P = 0.005). Among the 65 patients with the HLA-DR13 allele, more large tumors were found than in patients without that allele (P = 0.012). Finally, allele frequencies were analyzed in the three groups according to COMS criteria. No differences in the distribution of the alleles could be observed. 

Location of the tumor was classified into two categories—those with and those without ciliary body involvement. Of all tumors, 214 were choroidal and lacked ciliary body involvement and 21 included a ciliary body component. Within these categories, HLA allele frequencies did not deviate. Mean survival of patients with tumors without ciliary body involvement was 69.9 (SD, ±47.3) months, whereas the mean survival of patients with tumors with ciliary body involvement was 40.4 (SD, ±25.2) months (P = 0.015; log rank test). In patients carrying the HLA-B60 allele, tumors with involvement of the ciliary body were found to be more common (P = 0.030). 

Cell type was analyzed only in tumors in which histologic confirmation was obtained (e.g., all enucleated tumors). Tumors were classified as pure spindle in 47 cases, pure epithelioid in 25 cases, and of mixed cell type in the remaining 65 tumors. Survival according to cell type was 61.4 (SD, ±57.7) months for pure spindle, 43.5 (SD, ±40.3) months for the epithelioid cell type, and 51.9 (SD, ±41.8) months for the mixed cell-type tumors (P = 0.043; log rank test). An increased amount of tumors was classified as pure spindle cell type (P = 0.006) in HLA-B35–positive patients compared with patients who did not carry that allele (further comparisons not shown). 

Discussion

In this study, we examined whether HLA class I and II alleles confer susceptibility to uveal melanoma or are related to specific tumor characteristics or survival. Recent data suggest that NK cells, not CTLs, are the most relevant defense against the development of metastasis for patients with uveal melanoma. ⁶ Essential for these cells is the ability to recognize antigens by killer cell immunoglobulin-like receptors (KIRs), which are expressed on NK cells and T cells. The inhibitory and activatory ligands are particular epitopes present on class I molecules, specifically Bw4 on HLA-B and the C1 and C2 epitopes on HLA-C. ²⁴ Thus, the presence of particular HLA-B/-C haplotypes determines the repertoire of inhibitory and activatory ligands that can be used by KIRs to modulate NK cell–mediated responses. In this study of 235 patients with uveal melanoma and 2440 healthy blood donors, we did not find any significant HLA correlations, in contrast to earlier studies. Increased frequency of HLA-A32 is reported by Bertrams et al. ¹³ in 81 patients with choroidal melanoma compared with 1000 controls and by Martinetti et al., ¹⁴ who also report an increase of HLA-A32 in patient predisposition to spindle cell melanoma. It should be noted that direct comparison of these studies is complicated, because not all studies correct for the numbers of alleles tested. In our study, the frequency of HLA-DQ4 was increased (6% in the patient group compared with 3% in the control group), and the frequency of HLA-DQ5 was decreased (28% compared to 35%) before correction but failed to be significant after correction for the number of HLA alleles tested. Recently, Metzelaar-Blok et al. (IOVS 2002;43:ARVO E-Abstract 2236) looked at HLA class I and II and major histocompatibility complex class I related gene A (MICA) frequencies in 159 and 168 patients, respectively, and found no associations between the occurrence of uveal melanoma and any of these antigens. Our present study on HLA and uveal melanoma in 235 patients and 2440 healthy blood donors contained the largest number of patients tested. Although this study shows that HLA antigens do not contribute to an increased genetic susceptibility to uveal melanoma, this does not exclude an important role for HLA antigens in immune surveillance against uveal melanoma and their metastases. 

Survival analyses were performed for tumor size, location, and cell type. Large tumors, according to the COMS criteria, were significantly associated with decreased survival (P = 0.005). In addition, ciliary body involvement and epithelioid cell type correlated with bad prognosis (P = 0.015 and P = 0.043, respectively). These findings correspond to most studies in the literature. Although we had previously reported on an association between HLA-B40 (HLA-B40 is now “split” into HLA-B60 and HLA-B61) and death, we did not find that association in this series. The associations we found between HLA-DR13 and large tumors, between HLA-B35 and spindle cell type, and between HLA-B60 and ciliary body involvement are reported here for the first time but were not significant after correction. The relation between decreased survival in patients with the presence of a HLA-B44 allele has not been previously reported. 

Conclusions

In this study of 235 patients with uveal melanoma and 2440 healthy blood donors, we did not find any significant correlations after correction for the number of comparisons for any of the HLA alleles tested. Our data show that HLA class I and II allele type and development of uveal melanoma are unrelated. New correlations between tumor size, cell type, and ciliary body involvement were observed but were not significant after correction for the number of alleles tested. 

 Supported by Dutch Cancer Society (KWF) Grant RUL 2001–2472.

 Submitted for publication August 23, 2005; revised October 5, 2005; accepted November 28, 2005.

 Disclosure: W. Maat, None; G. W. Haasnoot, None; F.H. J. Claas, None; N. E. Schalij-Delfos, None; G.M. T. Schreuder, None; M. J. Jager, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Distribution of HLA-A and -B Antigens in Patients with Uveal Melanoma and Healthy Blood Donor

HLA Allele	Patients (n = 235)			Donors (n = 2440)
		No. Positive	% Positive	No. Positive	% Positive
A1	76	33	747	31
A2	131	57	1284	53
A3	55	24	700	29
A11	25	11	281	12
A23	4	2	60	2
A24	42	18	403	17
A25	6	3	46	2
A26	14	6	107	4
A28	17	7	244	10
A29	10	4	119	5
A30	3	1	85	3
A31	15	7	146	6
A32	11	5	149	6
A33	6	3	32	1
A66	1	0	11	0
B7	59	26	668	27
B8	62	27	554	23
B13	12	5	109	4
B14	6	3	70	3
B18	17	7	158	6
B27	16	7	157	6
B35	37	16	429	18
B37	5	2	99	4
B38	11	5	98	4
B39	13	6	79	3
B41	3	1	28	1
B44	55	24	586	24
B45	2	1	36	1
B47	2	1	12	0
B49	2	1	25	1
B50	1	0	29	1
B51	24	10	274	11
B52	1	0	15	1
B53	1	0	19	1
B55	9	4	105	4
B56	4	2	33	1
B57	19	8	133	6
B58	2	1	26	1
B60	29	13	361	15
B61	8	3	73	3
B62	39	17	370	15
B63	3	1	12	0
B70	1	0	30	1

  P values were corrected for 15 (HLA-A) and 28 (HLA-B) HLA antigens using the method of Edwards. All values were not significant.

Table 2.

 

Distribution of HLA-DR and -DQ Antigens in Patients with Uveal Melanoma and Healthy Blood Donors

HLA Allele	Patients (n = 235)			Donors (n = 2440)
		No. Positive	% Positive	No. Positive	% Positive
DR1	44	19	473	20
DR3	62	27	599	25
DR4	74	32	679	28
DR7	49	21	459	19
DR8	15	6	128	5
DR9	3	1	58	2
DR10	6	3	100	4
DR11	29	13	340	14
DR12	5	2	108	5
DR13	65	28	669	28
DR14	13	6	127	5
DR15	70	30	414	26
DR16	3	1	43	2
DQ2	90	39	881	37
DQ4	15	6	29	3
DQ5	66	28	300	35
DQ6	118	51	453	50
DQ7	58	25	652	28
DQ8	47	20	184	20
DQ9	21	9	71	8

  P values were corrected for 13 (HLA-DR) and 7 (HLA-DQ) HLA antigens using the method of Edwards. All values were not significant.

Table 3.

 

Tumor Characteristics from 235 (Tumor Size and Location) Patients with Uveal Melanoma and 137 with Enucleated Uveal Melanoma (cell type)

	n	%	Survival (mo)	SD	P
Tumor size
Small	31	13	86.4	41.8	0.005
Medium	155	66	70.9	48.0
Large	49	21	43.5	34.7
Location
Ciliary body involvement	21	9	40.4	25.2	0.015
No ciliary body involvement	214	91	69.9	47.3
Cell type
Spindle	47	34	61.4	57.7	0.043
Mixed	65	47	51.9	41.8
Epitheloid	25	18	43.5	40.3

Figure 1.

 

Kaplan-Meier analysis and log rank test showed the difference in survival according to presence or absence of the HLA-B44 allele (P = 0.012).

The authors thank Willem Verduyn (Department of Immunohaematology and Blood Transfusion, Leiden University Medical Center, Leiden, The Netherlands) and all the patients and controls for participating in this study.