Effectiveness of 1α-Hydroxyvitamin D2 in InhibitingTumor Growth in a Murine Transgenic Pigmented Ocular Tumor Model

Objective  To study the effectiveness of the vitamin D analogue 1α-hydroxyvitaminD₂ (1α-OH-D₂) in inhibiting ocular tumor growthin transgenic "Tyr-Tag" mice that developed pigmented ocular tumors producedwith the simian virus 40 T and t antigens under the control of the mouse tyrosinasegene. These animals develop pigmented intraocular tumors primarily from theretinal pigment epithelium that closely resemble the histologic features andgrowth pattern of human choroidal melanoma.

Methods  A total of 73 Tyr-Tag transgenic mice between 6 and 7 weeks old wererandomly assigned by sex and litter to 3 treatment groups to receive 0.05µg/d, 0.1 µg/d, or 0.2 µg/d of 1α-OH-D₂;a control group received vehicle (coconut oil). The drug was administeredby oral gavage 5 times a week for 5 weeks. The animals were then euthanizedand their eyes were enucleated and processed histologically. Three serialsections from each eye were examined microscopically and the mean tumor areameasured using Optimus software version 6.5 (Media Cybernetics LP, SilverSpring, Md). Toxic adverse effects were assessed on the basis of mortality,weight loss, and serum calcium levels.

Results  The mean tumor size in the 0.1-µg/d and 0.2-µg/d dose groupswas smaller than in the controls (P<.001). Nosignificant difference was seen between the 0.05-µg/d dose group andthe control group (P = .64). Survival for the 0.1-µg/dand 0.2-µg/d dose groups was lower than for the controls (95% in thecontrols vs 85.7% and 73.7%, respectively; P<.01).

Conclusion  In the Tyr-Tag transgenic mouse, 1α-OH-D₂ inhibitspigmented ocular tumor growth at moderate drug levels with relatively lowmortality.

Clinical Relevance  Vitamin D analogues merit further preclinical study in the treatmentof ocular melanoma.

In a recent "Perspective" on changing concepts in the treatment of choroidalmelanoma, Robertson concluded, "Gains in our ability to manage choroidal melanomawill likely be modest at best until effective systemic therapies can be identified."^1(p161) The present study was carried outto evaluate the effectiveness of a vitamin D analogue, 1α-hydroxyvitaminD₂ (1α-OH-D₂), in inhibiting tumor growth in atransgenic mouse ocular pigmented tumor model.

The role of vitamin D as a regulator of calcium and phosphate metabolismhas been known since early in the 20th century.² Morerecently it has been demonstrated that vitamin D analogues have importantbiological actions besides those related to mineral metabolism.^3-6 Theclassic signaling pathway, as with other steroid molecules, is through a nuclearreceptor, the vitamin D receptor that is a transcription factor. Studies duringthe past 20 years have shown vitamin D receptors in a wide range of tissues,⁷ as well as in numerous types of malignant cells.^8-10 Vitamin D analogueshave been shown to inhibit cellular proliferation, induce the differentiationof both normal and malignant cells, act as an antiangiogenic agent, and causeapoptosis.^11-17 Cutaneousmelanoma cells were one of the first malignant cells in which the antiproliferativeand prodifferentiation effects of vitamin D compounds were demonstrated.^18-20 Calcitriol and variousanalogues have also been demonstrated to significantly enhance the antitumorefficacy of other anticancer drugs in vitro and in vivo.^11,21,22 Theclinical potential of older agents has been limited by the induction of hypercalcemia.However, recent analogues have been developed with decreased calcemic activity,but potent antineoplastic and differentiating activity; these analogues havetherapeutic promise.²³ One such compound, 1α-OH-D₂, was used in the present study.

All research using mouse models of melanoma conform to the guidelinesset by the Research Animal Resources Center, University of Wisconsin, Madison,and the Association of Vision Research in Ophthalmology statement on the Useof Animals in Ophthalmic and Vision Research.

Pure crystalline 1α-OH-D₂ (Bone Care International,Middleton, Wis) was prepared for administration and the drug concentrationswere confirmed in the manner previously described.²⁴ Asolution of the drug was diluted in coconut oil to concentrations of 0.05µg per 0.1 mL; 0.1 µg per 0.1 mL; and 0.2 µg per 0.1 mL.The effectiveness and toxicity of these concentrations have previously beendescribed in dose-response studies in retinoblastoma animal models.^23-25

A total of 73 Tyr-Tag transgenic mice²⁶ between6 and 7 weeks old were randomly assigned by sex and litter to 1 of 3 1α-OH-D₂ treatment groups: (1) those that received 0.05 µg/d (approximately2.5 µg/kg), (2) those that received 0.1 µg/d (approximately 5.0µg/kg), and (3) those that received 0.2 µg/d (approximately 10µg/kg). In a serial study of the development of the ocular tumors, allmice were found to have small tumors at this age.²⁶ Acontrol group received 0.1 mL of vehicle (coconut oil). The treatment wasadministered by oral gavage 5 times a week for 5 weeks. The mice were maintainedon a vitamin D and calcium–restricted diet (vitamin D deficient andcalcium deficient, catalog 23980 PD; Purina Mills Inc, St Louis, Mo). Bodyweights were measured twice a week and just prior to euthanization on thelast day of treatment.

The mice were euthanized on day 35 of treatment. Their eyes were thenenucleated and placed in a 10% neutral-buffered formalin solution. Four seriallysectioned 5-µm-thick sections were cut from each of the superior, middle,and inferior areas of the globe in the manner previously described^23,24,27 and stained withhematoxylin-eosin. All 4 of the sections from each globe area were examinedunder a microscope and the section with the largest area of tumor was usedfor measurement. The outline of the tumor was traced on a microscopicallydigitized image and the tumor area measured using Optimus software version6.5 (Media Cybernetics, Silver Spring, Md). Three measurements from each tumorrepresentation were averaged to obtain the mean tumor measurement. These methodshave been described elsewhere.^23,24,27

Toxicity assessment for 1α-OH-D₂ was based on analysesof survival, changes in body weight, and serum calcium levels. Mice that diedprior to the end of the study were excluded from the body weight, and serumcalcium level. Autopsies were performed on all mice.

Serum calcium levels were measured in representative samples of micefrom each group selected at random (control group, 7 samples; 0.05-µg/ddose group, 5 samples; 0.1-µg/d dose group, 6 samples; and 0.2-µg/ddose group, 5 samples). Blood was drawn from axillary veins just prior toeuthanization, and calcium levels were analyzed by an independent commericallaboratory (Marshfield Laboratories Inc, Marshfield, Wis).

The effect of drug dose on tumor area, animal weight, and serum calciumlevel was assessed using 1-way analysis of variance. The tumor area was transformedto the log scale before calculating the mean area. Serum calcium data werenot transformed. Differences were considered statistically significant at P<.05.

The mean tumor area (×10⁴ µm²) ofthe various groups was as follows: control, 219.77; 0.05 µg/d of 1α-OH-D₂, 200.71; 0.1 µg/d of 1α-OH-D₂, 87.67; and 0.2µg/d of 1α-OH-D₂, 63.40 (Table 1 and Figure 1).A dose-dependent curve is apparent with higher-dose groups exhibiting a greaterdegree of tumor growth inhibition. The 0.1-µg/d and 0.2-µg/d dosegroups both showed significant tumor growth inhibition compared with the controlgroup (P<.001) and with the 0.05-µg/d dosegroup (P<.0011 and .0001, respectively). No primarycutaneous tumors or metastases were found in the control or treatment groups.

The survival rates for different treatment groups are shown in Table 1 and Figure 2. The mortality was significant in the 0.1- and 0.2-µg/ddose groups compared with the control group (P<.01for both groups).

The serum calcium levels increased with the increased dose of 1α-OH-D₂ (Table 1). The differencewas significant for the 0.1- and 0.2-µg/d dose groups compared withthe control group (P<.003 and P<.001, respectively). The 0.2-µg/d dose group was also significantcompared with the 0.05-µg/d dose group (P =.008)

There was no significant difference in mean body weight change—measuredas the difference between body weight at the start and at the end of the study—betweenvarious treatment groups. There was no significant difference between the0.05-µg/d dose group and the control group for any of the factors analyzed.

The incidence of melanoma in the white US population has increased from1 case per 100 000 in 1935 to 15 per 100 000 in 1996.²⁸ Theskin is the most common site of melanoma development; the eye is the secondmost likely site, with ocular melanoma constituting the most common primaryintraocular malignancy in adults. In the white population, ocular melanomahas an average annual incidence of 6 cases per 1 million with approximately1200 cases diagnosed each year in the United States.^29,30 Enucleationand radiation are the principal means of treatment. The recent CollaborativeOcular Melanoma Study (COMS) demonstrated that enucleation of large choroidaltumors, the principal method of treatment, was made neither more nor lesseffective when preceded by external irradiation.³¹ Themedium-sized tumor trial of COMS demonstrated that mortality rates followingbrachytheraphy were similar to mortality rates following enucleation for upto 12 years after treatment.³² Robertson¹ points out that in large tumors the estimated 5-yeartumor-specific mortality rate is approximately 33% for large tumors and 10%for medium-sized melanomas, and continues to increase when followed for upto 20 years. In addition, no effective treatment has been demonstrated formetastatic uveal melanoma. Although there is a trend toward earlier treatmentof small melanomas, there continues to be controversy regarding the indicationsfor treatment as well as the choice of specific therapy.¹ Clearlythere is a need for improved methods of treatment.

Much more information is available about the genetics and response totreatment of cutaneous melanoma than for ocular melanoma, but while thesetumors have many similarities, they also have some significant differences.³³ The genetics of uveal melanoma are less well definedthan cutaneous melanoma, and it appears that p16 (CDKN2A) mutations do not play a role in ocular melanomaas they do in certain cases of familial cutaneous melanoma.³³ Inaddition, patients with systemic ocular melanoma respond less well to chemotherapyand immunotherapy than do patients with systemic cutaneous melanoma.^34,35

A major deficiency in attempting to develop new methods of therapy forhuman uveal melanoma has been the absence of an ideal model animal for thehuman condition.³⁶ Although several animalmodels of human uveal melanoma have been used, each of these models has uniqueadvantages and disadvantages.³⁶ Spontaneousuveal melanoma rarely occurs in other species and its occurrence is unpredictable.Chemical- or irradiation-induced intraocular pigmented tumors may originatefrom the retinal pigment epithelium. Both feline leukemia–sarcoma virusand simian virus 40–induced uveal tumors failed to metastasize. TheGreene hamster melanoma and the B16 murine melanoma cells lines are virus-containingtumors that grow more aggressively than human uveal melanomas. Their biologicalbehavior after intraocular injection into various animal species is, however,predictable, allowing study of the mechanisms of growth and metastasis. Humancell lines can be grown in the eyes of immune-deficient mice, but the lackof an intact immune system is a significant artifact. Transgenic murine modelshave been developed using the promoter region of the tyrosinase gene to targetexpression of oncogenes in melanin-producing cells.^26,37-42 Spontaneousintraocular pigmented tumors and distant metastases may occur, although thesetumors develop primarily from the retinal pigment epithelium.

In this study we used Tyr-Tag transgenic mice that were produced withthe simian virus 40T and antigens under the control of the mouse tyrosinasegene.²⁶ These tumors were studied sequentiallyand found to develop primarily from retinal pigment epithelium, but theirgrowth and histologic appearance closely resemble that of human choroidalmelanoma.²⁶ Tumor development begins betweenbirth and 1 week of age, with lateral infiltration along Bruch's membranenoted at 6 weeks, choroidal invasion beginning at 7 weeks, and retinal invasionby 10 weeks.²⁶ Metastases occurred between12 and 40 weeks with the most common sites being subcutaneous tissues, lungs,retroperitoneal space, and brain.²⁶ The immunohistochemistryand electron microscopic features of these tumors were studied in detail,as was their response to chemotherapy and irradiation.²⁶

The antiproliferative-prodifferentiation effects of vitamin D compoundshave been shown in cultured human melanocytes, human melanoma cells, and melanomaxenograft models.^18-20 AMEDLINE search and review of the literature between 1982 and 2004 revealedmore than 40 additional articles regarding the response of cutaneous melanomato vitamin D. The literature indicates that all human melanoma cell linestested express the vitamin D receptor. This research has been carried outalmost exclusively with calcitriol and has been limited to preclinical studiesby the toxic effects associated with the induced hypercalcemia.

In the past several years, new vitamin D analogues have been producedwith reduced calcium effect but with equivalent or enhanced antineoplasticeffect compared with calcitriol or vitamin D₂.²³ Thecompound studied in the present experiments was 1α-OH-D₂.This has been demonstrated to induce low levels of hypercalcemia while providingeffective systemic serum drug levels for tumor treatment.^43,44 AnInvestigational New Drug application for 1α-OH-D₂ as a cancertreatment was submitted to the Food and Drug Administration in 1996. 1α-HydroxyvitaminD₂ has received Food and Drug Administration approval in both oraland intravenous formulations for the treatment of elevated parathyroid hormonelevels secondary to renal failure. In phase 1 trials for treatment of prostatecancer, 1α-OH-D₂ caused tumor growth suppression or stabilizationin some patients with low but reversible toxic adverse effects.⁴⁵ Thishas progressed to a phase 2 study.⁴⁶ This compoundalso proved effective in in vivo studies of treatment of retinoblastoma inthe LHbeta-Tag transgenic mouse.^24,25 Themechanism of action was demonstrated to be apoptosis associated with the up-regulationof both p53 and p2147 andinhibition of angiogensis.^47,48

In the present study 1α-OH-D₂ produced a statisticallysignificant decrease in intraocular tumor size in the 0.1- and 0.2-µg/ddoses. A reduction in tumor area was also seen with the 0.05-µg/d dose,but this was not statistically significant. As in previous experiments, theeffective doses produce a significant elevation in the level of serum calciumwith associated decreased survival.

These results suggest that further investigation of vitamin D analoguesas a possible therapy of melanoma is warranted. Its usefulness as a therapeuticagent both given individually and in combination therapy should be examinedin xenograft models using human melanoma cell lines as well as in other animalmodels for this tumor.

Correspondence: Daniel M. Albert, MD, MS, Department of Ophthalmologyand Visual Sciences, F4/344 Clinical Science Center, 600 Highland Ave, Madison,WI 53792-3284 (dalbert@wisc.edu).

Submitted for publication November 13, 2003; final revision receivedJanuary 30, 2004; accepted February 2, 2004.

This study was supported by grant EY01917 from the National Eye Institute,Bethesda, Md; and an unrestricted grant from Research to Prevent Blindness,New York, NY.

We thank Bone Care International for providing the pure crystalline1a-OH-D₂.

We acknowledge the assistance of Daniel Dawson, MD, and Joel Gleiser,MD, in treating the animals and measuring tumor size. Kirsten Hope contributedto preparation of the manuscript.