Correlation between vitiligo occurrence and clinical benefit in advanced melanoma patients treated with nivolumab: A multi‐institutional retrospective study
Abstract
Vitiligo is occasionally seen in melanoma patients. Although several studies indicate a correlation between vitiligo occurrence and clinical response in melanoma patients receiving immunotherapy, most studies have included heterogeneous patient and treatment settings. The aim of this study is to investigate the correlation between the occurrence of vitiligo and clinical benefit of nivolumab treatment in advanced melanoma patients. We retrospectively reviewed unresectable stage III or IV melanoma patients treated with nivolumab. Of 35 melanoma patients treated with nivolumab, 25.7% (9/35) developed vitiligo during treatment. The time from the start of nivolumab treatment to occurrence of vitiligo ranged 2–9 months (mean, 5.2). Of nine patients who developed vitiligo, two (22.2%) had a complete response to nivolumab and two (22.2%) had a partial response. The objective response rate was significantly higher in patients with vitiligo than in patients without vitiligo (4/9 [44.4%] vs 2/26 [7.7%]; P = 0.027). The mean time to vitiligo occurrence in patients achieving an objective response was significantly less than that in patients who showed no response (3.1 vs 6.8 months, P = 0.004). Vitiligo occurrence was significantly associated with prolonged progression‐free and overall survival (hazard ratio, 0.24 and 0.16; 95% confidence interval, 0.11–0.55 and 0.03–0.79; P = 0.005, and 0.047, respectively). At the 20‐week landmark analysis, however, vitiligo was not associated with a statistically significant overall survival benefit (P = 0.28). The occurrence of vitiligo during nivolumab treatment may be correlated with favorable clinical outcome.
Introduction
Vitiligo is associated with pigment loss of the skin due to the reduction or loss of functional epidermal melanocytes with T‐cell infiltration.1 Several studies indicate that vitiligo occurs in melanoma patients spontaneously or during treatment; its incidence rate ranges 2.8–43%,2-5 which is approximately 10‐fold higher than that in the general population.6 Although the detailed mechanism of vitiligo development in melanoma patients is still unclear, cellular and humoral immune responses are considered to play a crucial role for its development.7 The infiltration of the same CD8 T‐cell clones in vitiligo areas and melanoma sites along with the circulation of antibodies against melanoma‐associated antigens shared by melanocytes and melanoma cells (such as tyrosinase‐related proteins [TRP]1 and 2, glycoprotein [gp]100, and melanoma antigen recognized by T‐cells [MART]‐1, each of which is correlated with melanin synthesis) suggest an autoimmune mechanism.6, 8-10
Furthermore, a clinical correlation between vitiligo occurrence and improved prognosis in melanoma patients has been reported in several studies.2, 5, 11 Richards et al. first reported a significant correlation between vitiligo occurrence and tumor shrinkage in patients treated with sequential chemoimmunotherapy, including interleukin‐2 and interferon (IFN)‐α‐2.12 Subsequently, there have been other immunotherapy studies regarding vitiligo occurrence in association with clinical response and prolonged survival in stage III and IV melanoma patients.3, 4, 13, 14 In addition, a recent systematic review and meta‐analysis reported by Teulings et al. also indicated that vitiligo occurrence was significantly associated with prolonged progression‐free and overall survival of stage III–IV melanoma patients in 27 immunotherapy studies, most of which were evaluating the clinical efficacy of vaccine therapies.15
In contrast, very few studies evaluating immune‐checkpoint inhibitor programmed death receptor‐1 (PD‐1) antibodies, such as pembrolizumab and nivolumab, suggest the potential relationship between vitiligo occurrence during treatment and clinical efficacy in advanced melanoma patients.16, 17 Furthermore, these studies include heterogeneous patient settings and treatment modalities, including adjuvant therapy and therapy for metastatic melanoma, different combination with peptide vaccines, and various doses in a study. Therefore, the purpose of this study was to investigate the occurrence of vitiligo in advanced melanoma patients treated with a fixed dose of nivolumab in order to assess the relation between its occurrence and clinical benefit of nivolumab.
Methods
This retrospective study included patients with advanced melanoma treated with nivolumab at the Department of Skin Oncology/Dermatology, Saitama Medical University International Medical Center and at the Department of Dermatology, University of Tsukuba between 1 January 2012 and 31 December 2015. Patients’ characteristics, including demographic, clinical, pathological, treatment and outcomes, were extracted from the patients’ medical records. The study was approved by the institutional review boards and human research ethics committees at each institution, and was performed in accordance with the ethical guidelines of the 1975 Declaration of Helsinki (2008 revision).
Patients
This study included patients with unresectable stage III or IV metastatic melanoma according to the 2009 American Joint Committee on Cancer melanoma staging and classification,18 who were previously treated with cytotoxic chemotherapy. Patients who had been enrolled in a phase II clinical trial of nivolumab conducted in Japan (ONO‐4538‐02) were also included in this study. Individuals who received only one cycle of nivolumab, whose follow‐up period was less than 1 month, and who had no radiological response assessment were excluded from the study.
Treatment
Patients were treated with i.v. nivolumab at a dose of 2 mg/kg of bodyweight every 3 weeks, which is the licensed dose and administration of nivolumab in Japan. The treatment was discontinued if the patient developed grade 3 or 4 toxicities, such as pneumonitis and type 1 diabetes. If the patient achieved a complete response (CR) or progressive disease (PD), the treatment discontinuation depended on the patient's request and careful discussion with the patients. Treatment was also discontinued when the patients who had been enrolled in a phase II study (ONO‐4535‐02) achieved CR or PD.
Vitiligo assessment
Patients received a whole‐body skin examination by a board‐certified dermatologist prior to nivolumab treatment and every 3 weeks during treatment. After the patients discontinued nivolumab treatment, the interval of skin examination was increased to 3–6 weeks. The areas of vitiligo were recorded as a photograph using a single lens reflex digital camera during every visit. We evaluated the time to vitiligo occurrence and number of cycles of nivolumab before vitiligo occurrence, based on the US National Cancer Institute's common toxicity criteria, version 4.03 (14 June 2010; http://evs.nci.nih.gov/ftp1/CTCAE/CTCAE_4.03_2010-06-14_QuickReference_8.5x11.pdf#search='CTCAE+version+4.03). Vitiligo, or skin hypopigmentation in these toxicity criteria, was graded in each patient.
Response evaluation
Clinical responses to nivolumab were assessed by computed tomography using Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1.19 The definition of responses is as follows: CR is the disappearance of all measurable disease; partial response (PR) is at least a 30% decrease in the sum of target lesions; PD is at least a 20% increase in the sum of target lesions or appearance of one or more new lesions; and stable disease (SD) is insufficient shrinkage to qualify for a PR and insufficient increase to qualify for PD.
Statistical analysis
The patients’ demographic and clinical characteristics were compared between the patients with and without vitiligo occurrence using Student's t‐test, Pearson χ2‐test or Fisher's exact test, as appropriate. Associations between the presence or absence of vitiligo and response rate were compared and assessed using the Pearson χ2‐test or Fisher's exact test. The relation between timing of vitiligo occurrence and clinical response was also evaluated using Student's t‐test. Kaplan–Meier analysis was used to estimate differences in overall survival, according to a stratified log–rank test. Overall survival analysis was also performed using this landmark assessment of vitiligo occurrence within 20 weeks after the start of nivolumab treatment (i.e. a patient alive and had vitiligo by week 20 was compared with a patient without vitiligo by week 20).
All data were analyzed using StatMate V (Atoms, Tokyo, Japan), and P < 0.05 was considered statistically significant.
Results
Patient characteristics
In total, 37 patients with unresectable stage III or stage IV melanoma were treated with nivolumab. Two patients were excluded from the study. Both patients received only one cycle of nivolumab owing to rapid disease progression and death within a month (one patient) or severe toxicity (occurrence of severe psoriasis). Therefore, the remaining 35 patients were enrolled in this study.
Vitiligo occurred in nine of 35 patients (25.7%) during nivolumab treatment. Mean time from the start of nivolumab treatment to vitiligo occurrence was 5.2 months (range, 2–9). The mean number of cycles of nivolumab before the vitiligo occurrence was 6.1 (range, 2–13). Eight of nine patients (89%) had a grade 1 vitiligo, with a total area of vitiligo of less than 10% of body surface area (Table 1).
| Characteristic | Clinical data |
|---|---|
| Incidence rate of vitiligo | 9/35 patients (25.7%) |
| Time to occurrence of vitiligo | 5.2 months (range, 2–9) |
| No. of cycles of nivolumab before occurrence of vitiligo | 6.1 (range, 2–13) |
Baseline characteristics showed no statistically significant difference between patients with vitiligo occurrence (vitiligo group) and patients without vitiligo occurrence (non‐vitiligo group) (Table 2). The mean age in the vitiligo group was 65.3 years (range, 40–79), and four patients (44.4%) were male and five patients (55.6%) were female. Each group had only one patient with unresectable stage III melanoma on the vagina (11.1% vs 3.9%), while the remaining patients in each group had stage IV metastatic melanoma. There was no relationship between vitiligo occurrence and the number of cycles of nivolumab (mean, 8.4 vs 7; P = 0.4). All patients had previously received cytotoxic chemotherapy, such as dacarbazine, vincristine or cisplatin. No patients received immunotherapy, except for adjuvant therapy prior to nivolumab treatment. In the non‐vitiligo group, one patient with a BRAF mutation was previously treated with vemurafenib monotherapy. In the vitiligo group, two patients had a history of limited vitiligo occurrence prior to nivolumab treatment (Table 2). One patient had a 2‐cm area of vitiligo on the abdominal skin overlying the single subcutaneous metastasis. The area of vitiligo had been excised together with this metastatic lesion 2 years before the nivolumab treatment. Since then, the patient had had no vitiligo development until she started nivolumab. The other patient had a 13‐year history of a 1‐cm area of persistent vitiligo on the jaw after she had received two cycles of systemic chemotherapy using dacarbazine, nimustine, vincristine and IFN‐β as adjuvant chemotherapy. This person also did not show additional vitiligo progression before starting nivolumab treatment.
| Characteristic | Vitiligo group | Non‐vitiligo group | P |
|---|---|---|---|
| Total no. | 9 | 26 | – |
| Age (years) | |||
| Mean | 65.3 | 68.7 | 0.45 |
| Range | 40–79 | 41–85 | |
| Sex | |||
| Male | 4 (44.4%) | 14 (53.8%) | 0.63 |
| Female | 5 (55.6%) | 12 (46.2%) | |
| Stage | |||
| Unresectable stage III | 1 (11.1%) | 1 (3.9%) | 0.42 |
| Stage IV | 8 (88.9%) | 25 (96.1%) | |
| Total cycle of nivolumab injection | 8.4 (2–17) | 7 (2–18) | 0.4 |
| History of cytotoxic chemotherapy | 9 (100%) | 26 (100%) | – |
| History of prior immunotherapy (excluding adjuvant therapies) | 0 (0%) | 0 (0%) | – |
| History of prior targeted therapy | 0 (0%) | 1 (3.84%) | 0.55 |
| History of prior vitiligo | 2 (22.2%) | 0 (0%) | 0.1 |
Response rate
Among the nine patients in the vitiligo group, two (22.2%) patients had a CR, two (22.2%) had a PR, five (55.6%) had an SD and no patients had a PD. Among the 26 patients in the non‐vitiligo group, one (3.8%) had a CR, one (3.8%) had a PR, 11 (42.3%) had a SD and 13 (50%) had a PD. An objective response rate of 44.4% in the vitiligo group and 7.6% of patients in the non‐vitiligo group was observed (P = 0.027) (Table 3).
| Patient group | Best response | ||||
|---|---|---|---|---|---|
| Complete | Partial | Stable | Progressive | Response rate | |
| Vitiligo, n (%) | 2 (22.2) | 2 (22.2) | 5 (55.6) | 0 (0) | 4 (44.4)aa
Fisher's exact test, P = 0.027.
|
| Non‐vitiligo, n (%) | 1 (3.8) | 1 (3.8) | 11 (42.3) | 13 (50) | 2 (7.6)aa
Fisher's exact test, P = 0.027.
|
| Total, n (%) | 3 (8.6) | 3 (8.6) | 16 (45.7) | 13 (37.1) | 6 (17.1) |
- a Fisher's exact test, P = 0.027.
Relation between time to vitiligo occurrence and response rate
The mean time from the start of nivolumab to vitiligo occurrence was 3.1 months in patients who showed an objective response and 6.8 months in patients who did not show response (P = 0.004). All patients who showed a tumor response developed vitiligo within 5 months (Fig. 1).
Progression‐free and overall survival
Vitiligo occurrence was significantly associated with progression‐free survival (hazard ratio [HR], 0.24; 95% confidence interval [CI], 0.11–0.55; P = 0.005) and overall survival (HR, 0.16; 95% CI, 0.03–0.79; P = 0.047) (Fig. 2a,b). At the 20‐week landmark analysis, however, vitiligo was not associated with a statistically significant overall survival benefit (P = 0.28) (Fig. 2c).
Discussion
In the present study, we evaluated the correlation between vitiligo occurrence and clinical efficacy of nivolumab for advanced melanoma. To our knowledge, there is only one other study investigating association between vitiligo and clinical benefit to nivolumab.16 However, this previous study was conducted in heterogeneous patient settings, with individuals who received nivolumab in combination with a peptide vaccine such as gp100, NY‐ESO‐1 or Melan‐A, as well as doses of 1 mg/kg or 3 mg/kg of nivolumab.16
Nivolumab is a fully human immunoglobulin G4 monoclonal antibody directed against the negative immunoregulatory cell surface receptor PD‐1. Nivolumab binds to and blocks the activation of PD‐1 by its ligands programmed cell death ligand 1 (PD‐L1), which is overexpressed on tumor cells. This leads to the activation of T cells and immune responses against tumor cells.20 Recent clinical trials of nivolumab for advanced melanoma have shown considerable clinical benefit,21, 22 which led to the approval of this drug worldwide. Nivolumab is well tolerated, and grade 3 or 4 treatment‐related adverse events in the phase III clinical trials account for approximately 11.7–16.3% of the patients. The frequent grade 1 or 2 adverse events reported have been fatigue (19.9–32.9%), rash (15–25.3%), pruritus (16.5–18.8%), diarrhea (15–17%) and nausea (13–16.5%), while the incidence rate of vitiligo has been reported to be 7.3–10.7%.22, 23
Vitiligo in association with melanoma patients treated with immunotherapy is considered to be correlated with clinical efficacy, as a manifestation of melanoma immunity.24, 25 In the PD‐L1/PD‐1 pathway, vitiligo likely mediates peripheral tolerance of melanosomal proteins such as tyrosinase and TRP‐2, and blocking of this pathway may induce autoimmune vitiligo.26 The overexpression of melanoma‐associated antigens and the release of these antigens in melanoma patients during immunotherapy leads to the breakdown of peripheral tolerance against these antigens,24 and PD‐1 antibody such as nivolumab blocks an inhibitory signal to melanoma cells and activates the immune response against antigens shared by melanocytes and melanomas. Indeed, vitiligo has not been reported as an adverse event of nivolumab in patients with renal cell cancer and lung cancer.27-29
Our data suggests that the occurrence of vitiligo is associated with a favorable clinical response in melanoma patients treated with nivolumab. The objective response rate in the current study was 44.4% in persons with vitiligo compared with 7.6% in persons who did not have vitiligo (P = 0.027). Additionally, 50% of patients without vitiligo developed PD and 12 patients died within 9 months. These data are quite similar to those in the recent prospective study reported by Hua et al., which reported on the association of vitiligo with tumor response after pembrolizumab treatment, although this prospective study includes patients treated with different doses of pembrolizumab (1, 3 and 10 mg/kg).17 This prospective study showed that an objective response was associated with a higher occurrence of vitiligo than no vitiligo occurrence (71% vs 28%, P = 0.002).
The mean time to vitiligo occurrence in the present study was approximately 5 months after the start of nivolumab treatment. This result indicates that we cannot regard the occurrence of vitiligo as an early marker of good clinical response, because the first radiological tumor assessment is usually performed earlier than 5 months after the start of nivolumab treatment. However, vitiligo occurrence within approximately 5 months in the present study was significantly correlated with favorable clinical response, which may be a prognostic factor of good tumor response to nivolumab. While recent studies have reported that the median time to onset of vitiligo after the start of nivolumab and pembrolizumab is 1.3 and 4.2 months, respectively,16, 17 an earlier mean time to vitiligo occurrence may reflect a synergistic effect of a peptide vaccine in combination with nivolumab in this study.16
The present study also showed the significant improvement in progression‐free and overall survival associated with vitiligo, although the 20‐week landmark analysis did not show a significant survival benefit in patients who developed vitiligo. Similar data are seen in pembrolizumab studies, demonstrating that overall survival in patients presenting with vitiligo during the first 12, 16 and 20 weeks was not significantly improved (P = 0.35, 0.17 and 0.17, respectively).17 In contrast, conflicting results are seen in the study of nivolumab, showing that vitiligo significantly enhances the survival benefit at 20‐week landmark analysis (P = 0.028).16 Thus, vitiligo occurrence during PD‐1 antibody treatment appears to represent a time‐dependent factor. When the correlation between the occurrence of vitiligo and survival is evaluated in melanoma patients, a guarantee‐time bias must be taken into account,30 because patients who developed vitiligo received treatment and lived long enough to present with vitiligo. To address this potential limitation, we performed a 20‐week landmark analysis, which excludes patients who died before that point, thus reducing sample size and power. After controlling for guarantee‐time bias in the landmark analysis, the correlation between the appearance of vitiligo and overall survival did not reach statistical significance. This exclusion is also considered to produce “exclusion bias” if the patient's survival and prognosis truly depends on the occurrence or absence of vitiligo during nivolumab treatment. Therefore, the patients without vitiligo occurrence may reflect scant immunoreactivity of nivolumab against the tumor cells that may lead to early death. Other limitations include that the sample size for this study was small and the follow‐up period was short.
In conclusion, the present study suggests that vitiligo occurrence may be associated with clinical response and survival in melanoma patients treated with nivolumab. Owing to the limitations described above, a large randomized prospective study with long‐term follow up is needed to clarify the mechanisms and relationship of nivolumab‐associated vitiligo and tumor response.
Acknowledgments
This work was partly supported by the National Cancer Center Research and Development Fund (26‐A‐4).
Conflict of Interest
Y. N. is on a scientific advisory board for Chugai Pharmaceutical Company. Y. F. is on scientific advisory boards for Ono and Chugai Pharmaceutical Companies.
