Vaccine-specific local T cell reactivity in immunotherapy-associated vitiligo in melanoma patients

Patient selection

Melanoma patients with regional lymph node and distant metastases, participated in an experimental study using autologous monocyte-derived DCs loaded with HLA-A2.1 compatible tumor antigens gp100 and tyrosinase [5]. Only patients who developed vitiligo after immunotherapy, clinically confirmed by a dermatologist, were selected. Blood samples and 6 mm biopsies of perilesional vitiligo were obtained after informed consent. Approval from the local regulatory committee was obtained.

Preparation of vitiligo-infiltrating lymphocytes (VILs)

Six mm punch biopsies of the perilesional vitiligo-site (Fig. 1a) were obtained and cut in half. One part was cryopreserved for immunohistochemistry at a later timepoint, the other part was disrupted and a cell suspension was made by gentle squeezing in a sterile open filter chamber (NPBI, Amsterdam, The Netherlands) in RPMI/7% human serum (Sanquin, Nijmegen, The Netherlands) supplemented with IL-2 (100 U/ml). The cell suspension was placed in a 24-wells plate (Costar Badhoevedorp, The Netherlands). T cells were morphologically and functionally tested after 2 weeks of culture.

Preparation of tumor-infiltrating lymphocytes (TILs) and melanoma cell lines

TILs were obtained and cultured according to the same protocol as the vitiligo cell suspension (see above). A tumor cell line was generated by tumor culture in DMEM (Gibco—Invitrogen) supplemented with 10% fetal calf serum (Greiner). Fresh medium was added twice a week.

MHC Tetramer staining

VILs and TILs (50 × 10³ cells in 10 μl) or 1 × 10⁶ peripheral blood mononuclear cells in 10 μl, were incubated with directly labeled gp100^154–162, gp100^280–288, tryrosinase^369–376 and Melan-A^26–35 and HIV^77–85 tetrameric-MHC-HLA-A2.1 complexes for 60 min at room temperature. FITC-conjugated monoclonal antibodies directed against CD8 (Becton Dickinson) were added during the last 20 min. After washing, the samples were analyzed by flow cytometry. Percentage tetramer positive cells were calculated over the CD8⁺ fraction. All samples were tested with HIV^77–85-HLA-A2.1-tetramers recognizing the irrelevant HIV-peptide SLYNTVATL as background control [5]. Control tetramer binding was less than 0.02% in all samples (supplementary figure 1).

Cytokine production

Production of cytokines by VILs and TILs were measured in response to specific stimuli (target:effector-ratio 1:1). Two different target cells were used: T2 cells pulsed with gp100^154–162, gp100^280–288, tryrosinase^369–376 or the irrelevant peptide HIV^77–85 and BLM cells (a melanoma cell line expressing HLA-A2.1 and no endogenous expression of gp100 and tyrosinase) transfected with control antigen G250, gp100 or tyrosinase. IFN-γ and IL-2 production were measured in supernatants after 16 hours by cytometric bead array according to the manufacturers protocol (Th1/Th2 Cytokine CBA 1; BD Pharmingen).

Antibodies and immunostaining

Regulatory T cell phenotype was assessed by flowcytometry using mAbs directly conjugated to fluorescent dyes. The FOXP3-FITC (clone PCH101, eBioscience, San Diago, CA, USA), CD4-APC, CD25-PE, CD127-PE-Cy5 (all Becton Dickinson) were used according to manufacturers protocols. Regulatory T cells were defined as CD4⁺FoxP3⁺CD25⁺CD127⁻.

For immunohistochemistry, the following mAbs were used: M2-7C10+M2-9E3 (Labvision, Fremont, CA, USA) against Melan-A, SP7 (Labvision) against CD3, HMB-45 (Dako, Glostrup, Denmark) against gp100 and T311 (Novocastra, Newcastle, UK) against tyrosinase [6].

Patients

We prospectively followed three HLA-A2.1 positive patients with histologically documented melanoma who received DC-based immunotherapy and subsequently developed vitiligo (Table 1).

Patient 1. Patient 1 was diagnosed, two years prior to inclusion, with melanoma of the scalp with regional lymph node metastases (T3aN2aM0), for which a radical lymph node dissection was performed. She relapsed with regional lymph node and cutaneous metastases, which were resected and treated with adjuvant radiotherapy. At inclusion, this patient presented with multiple subcutaneous and lung metastases. The patient was vaccinated intranodally with antigen loaded mature DCs [5] but treatment was discontinued after 3 vaccinations because of progression of cutaneous metastases. She developed vitiligo in the flank during further tumor progression.

Patient 2. Patient 2 was diagnosed, twelve years prior to inclusion, with melanoma located on the right lower leg (T1aN0M0), which was surgically removed. From 1997 on, the patient relapsed with in transit and distant cutaneous metastases, for which he was treated with regional limb perfusion (melfalan) and dacarbazine. At inclusion, he was diagnosed with lymph node metastases and multiple (sub-)cutaneous metastases for which he was repeatedly vaccinated intranodally with antigen loaded mature DCs. After three vaccinations, the patient developed vitiligo in the neck region and dorsal side of both hands. To date, the vitiligo still progresses and affects the head, back, hands and legs.

Patient 3. Patient 3 was diagnosed, 6 months prior to inclusion, with a superficial spreading melanoma of the left thigh (T1a/bN0M0). At inclusion she presented with an inguinal lymph node metastasis, for which a radical lymph node dissection was performed. She was vaccinated intradermally in an adjuvant setting with antigen loaded mature DCs. After eight vaccinations, she developed a severe rash in the neck region, thorax, back and upper extremities and subsequent vitiligo in these areas.

Vitiligo development in melanoma patients upon vaccination with DCs has been reported before [15, 24]. Although not investigated, it is currently believed that the vitiligo in these patients results from sensitization to antigens shared by melanocytes and melanoma cells. To address this question we took perilesional biopsies and studied the vitiligo infiltrating immune cells.

Vaccine specific lymphocytes infiltrate vitiligo lesions

Immunohistochemical analysis of the vitiligo in all three patients demonstrates perilesional infiltration of T cells (Fig. 1a, b). With tetramer analysis, CD8⁺ T cells specific for one or more gp100 and tyrosinase HLA-A2.1 epitopes were detected at low frequencies in the peripheral blood after vaccination in all three patients. Gp100 or tyrosinase specific CD8⁺ T cells were not detectable prior to vaccination (data not shown). Interestingly, specific CTLs were detected at increased concentrations in the vitiligo lesion compared to peripheral blood (Table 1; supplementary figure 1). This phenomenon was most obvious in patient 3 (Fig. 1c, d). The accumulation of these CTLs in the vitiligo lesions can result from specific migration [26] and/or a local proliferation advantage for MDA specific CTLs induced by perilesional antigen exposure.

Although the immunotherapy was not primarily directed against the Melan-A epitope, Melan-A specific CTLs were observed in the vitiligo lesions and tumor lesion of patient 1 and 2 (Table 1; Fig. 3). This finding is consistent with the observation that anti-tumor vaccines can have effects beyond their intrinsic specificity. It is shown that the interaction of vaccine specific T cells with melanoma cells may trigger a broad activation of other anti-tumor T cells, a phenomenon called epitope spreading [10, 14].

Vitiligo infiltrating lymphocytes produce IFN-γ after specific stimulation

To test whether vitiligo infiltrating lymphocytes (VILs) were functional, we stimulated these cells with gp100 or tyrosinase expressing target cells and measured subsequent IFN-γ and IL-2 production. VILs of patients 1 and 3 produce IFN-γ and IL-2 when exposed to gp100-target cells and VILs of patient 2 produce IFN-γ and IL-2 when exposed to both gp100 and tyrosinase-target cells. No cytokines were produced when VILs were exposed to unloaded control target cells (Fig. 2).

From patients 1 and 2 we were able to culture melanoma cell lines from metastases. Interestingly, the VILs also produced IFN-γ and IL-2 when stimulated by the autologous tumor cell line (Fig. 2). The cross reactivity of the VILs with autologous tumor cells further supports the hypothesis that the vitiligo is a marker of immunity against melanoma cells.

Vitiligo lesions in our patients did not specifically occur at the vaccine injection-site, at the site of the primary tumor, or in close proximity of new metastases. From this we conclude that the melanoma-associated vitiligo is not caused by a direct and local side-effect of tumor destruction, but rather by the presence of circulating MDA-specific CTLs. Although we did not use MHC class II peptides in our vaccination protocol, our data do not exclude that specific CD4⁺ cells are involved. MDA-specific epitopes recognized by CD4⁺ T cells have been reported [31] and it is also known that melanocytes can present peptides in an MHC class II restricted manner [28]. These observations suggest that melanocyte destruction may not only depend on MHC class I-restricted cells.

Immunotherapy-associated vitiligo and prognosis

Our case series is too limited to afford decisive conclusions on the prognostic value of vitiligo on melanoma disease. The presence of durable non-progressive disease in patients 2 and 3 suggest that the immunotherapy induced vitiligo lesions are a favourable prognostic factor. We further show that all three patients have a relatively high number of MDA-specific T cells in the blood, DTH biopsies and melanoma. Earlier papers have described a correlation with the presence of MDA-specific T cells in these compartments and favourable outcome [2, 5].

Despite the presence of functional melanoma specific T cells, patient 1 had progressive disease. This is intriguing since the tumor expressed the MDAs gp100 and Melan-A (IHC, Fig. 3a, b) and was infiltrated by CTLs that specifically recognize these epitopes (inserts Fig. 3a, b; Table 1). Several mechanisms may account for the escape of melanoma cells to immune surveillance [21]. It has recently been published that CD4⁺FoxP3⁺CD25⁺ regulatory T cells (Tregs) can infiltrate melanoma tumors and locally suppress cytotoxic anti-tumor responses, also in vaccinated melanoma patients [1]. Infiltration of melanoma tumors with Tregs correlates with poor prognosis [27]. In the melanoma metastasis of patient 1, we detected that 27% of the tumor infiltrating CD4⁺ T cells consisted of CD4⁺FoxP3⁺CD25⁺CD127^- Tregs vs 4% of Tregs in the blood of this patient (data not shown).

Altogether, our data support an association between vitiligo and favourable outcome. However, active vaccine-induced vitiligo does not exclude tumor progression. The dichotomy between MDA-specific CTL responses in melanoma and vitiligo might result from quantitative CTL differences, the quality of the CTLs such as T cell receptor affinity and cytokine production [20], the amount of antigen presented [8, 23] and the different environmental conditions in which these T cells exert their function [21].