Clinical Report

Accepted Jul 5, 2017; Epub ahead of print Jul 6, 2017

Acta Derm Venereol 2017; 97: xx–xx.

Corr: Ryuhei Okuyama, Department of Dermatology, Shinshu University School of Medicine, 3-1-1 Asahi, Matsumoto 390-8621, Japan. E-mail: rokuyama@shinshu-u.ac.jp

In the advanced stages, melanoma is among the most aggressive and treatment-resistant cancers and its incidence is increasing worldwide. Substantial progress has been made recently in the treatment of advanced melanoma. Immune checkpoint blockade is recommended as the first-line therapy for patients with metastatic or unresectable melanoma. Monoclonal antibodies against programmed cell death-1 (PD-1) and cytotoxic T-lymphocyte-associated protein-4 (CTLA-4) showed remarkable long-term benefits in 40% and 10% of patients, respectively (1, 2). These antibodies act directly on the immune system, rather than on the tumour, and the kinetics of tumour regression may be delayed. In some cases, tumours assessed using computed tomography (CT) imaging appear to enlarge during therapy before regressing. In other patients, new tumours appear during therapy and later regress. The apparent tumour progression may, in some cases, reflect intratumoural inflammation rather than actual tumour growth. Despite considerable progress in the treatment of melanoma, reliable markers to predict treatment response or evaluate early recurrence are lacking. Non-invasive indicators of changes in tumour burden could provide early information about potential therapeutic outcomes, thereby preventing potentially serious immune-related adverse events in patients with true disease progression.

Serum lactate dehydrogenase (LDH) is widely used as a surrogate marker of tumour progression in melanoma (3). Elevated LDH level is associated with a high disease burden and poor survival. However, because LDH occasionally remains within the normal range despite tumour progression, its value as a biomarker is limited. In addition, LDH increases in response to many conditions, such as infections, liver dysfunction, and myocardial disorders among others, and in malignancies other than melanoma.

Circulating tumour DNA (ctDNA) in the peripheral blood is a novel biomarker for evaluating tumour features in patients with advanced cancer. Cell-free DNA is released from both normal and tumour cells via a variety of mechanisms, including apoptosis and necrosis. ctDNA contains the same genetic alterations present in the source tumour (4–6). In particular, ctDNA with mutations in EGFR, KRAS, or BRAF is detected in cancer patients, and the detection of mutations in ctDNA is under investigation as a specific biomarker for the diagnosis and monitoring of patients with different cancer types (7–11). Mutations of BRAF and NRAS are common in melanoma: the frequency of BRAF mutations, chiefly BRAF^V600E, is 50–60% in Caucasians and 20–30% in Asian populations (12–14); and that of NRAS mutations is 15–25% in Caucasians and 7–10% in Asian populations (14–16). Therefore, BRAF-mutated and NRAS-mutated ctDNAs are potentially useful biomarkers for melanoma.

In the present study, dynamic changes in ctDNA were analysed in parallel with measurement of LDH levels in melanoma patients during treatment with anti-PD-1 antibody. Droplet digital PCR (ddPCR) technology, which can be readily used to quantify mutant DNA copies, was used to examine changes in mutant ctDNA levels across different time-points. The results indicated that BRAF-mutated and NRAS-mutated ctDNAs are potentially useful biomarkers of treatment response to anti-PD-1 immunotherapy in patients with melanoma.

Four melanoma cell lines were used in this study: Mel-2 (BRAF^wild-type, NRAS^wild-type), A375 (BRAF^V600E, NRAS^wild-type), WM3282 (BRAF^V600K, NRAS^wild-type) and WM3406 (BRAF^wild-type, NRAS^Q61K). Mel-2 was established from Japanese patients, and the other lines were established from Caucasian patients. Mel-2 was obtained from Professor Y. Kawakami (Keio University School of Medicine, Tokyo). A375 was obtained from Professor T. Nagatani (Tokyo Medical University School of Medicine, Tokyo). Mel-2 and A375 were cultured in RPMI-1640 medium supplemented with 10% heat-inactivated foetal bovine serum and 1% penicillin-streptomycin at 37°C in a humidified atmosphere of 5% CO₂. WM3282 and WM3406 cells were purchased from Rockland (Limerick, PA, USA) and cultured in 80% MCDB153 and 20% Leibovitz’s L-15 supplemented with 2% heat-inactivated foetal bovine serum and 1.68 mM CaCl₂ at 36°C in a humidified atmosphere of 5% CO₂. Cells were fed twice per week and passaged by exposure to 0.25% trypsin prepared in ethylenediaminetetraacetic acid (EDTA).

Five patients with BRAF-mutated and NRAS-mutated melanoma who had metastatic disease were evaluated. Patients were treated with the anti-PD-1 antibody nivolumab at 2 mg/kg every 3 weeks. Nivolumab was followed by vemurafenib (BRAF inhibitor) 1,920 mg/day and ipilimumab (anti-CTLA-4 antibody) at 3 mg/kg every 3 weeks in Patients 1 and 2, respectively. CT imaging was performed at 2- or 3-month intervals, and tumour responses to treatment were evaluated using Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 (CR, complete response; PR, partial response; SD, stable disease; and PD, progressive disease) (17). Patient characteristics are summarized in Table I. The patients were treated in the Department of Dermatology of Shinshu University Hospital. Diagnosis of melanoma was patho-logically confirmed by haematoxylin and eosin staining, and the presence of mutations was confirmed in tumour tissues. In brief, genomic DNA was extracted from formalin-fixed, paraffin-embedded tumour specimens. Exon 15 of BRAF and exons 2/3 of NRAS, which include mutational hot spots, were amplified by PCR and sequenced (12, 16). All melanoma patients enrolled in the study provided written informed consent for the use of their peripheral blood and resected tumour tissues. Twenty-eight healthy subjects were studied as control samples. Previous data on the BRAF^V600K signal in the peripheral blood of healthy subjects were also included in the analysis (18). The study was approved by the Ethics Committee of the Shinshu University School of Medicine and conducted according to Institutional Review Board guidelines.

Blood samples (7 ml) were collected in tubes containing EDTA. Within 2 h after collection, blood samples were centrifuged at 1,900 g for 10 min, followed by a further 10 min of centrifugation at 16,000 g to separate plasma, and then stored at −80°C until analysis. Cell-free DNA was extracted from 2 ml of plasma using the QIAamp Circulating Nucleic Acid Kit (Qiagen, Tokyo). BRAF-mutated and NRAS-mutated ctDNAs were quantified by PrimePCR™ ddPCR™ mutation assay kit (Bio-Rad, Hercules, CA, USA: For manufacturer’s information, see http://www.bio-rad.com/en-jp/product/primepcr-pcr-primers-assays-arrays). The reaction mixture (20 µl) contained ddPCR Supermix (Bio-Rad), amplification primer/probe mix specific for BRAF or NRAS (Bio-Rad), template DNA, and water. Droplets were generated from 20 µl of reaction mixture using a QX200 Droplet Generator (Bio-Rad). Next, droplets were transferred to PCR plates, and the PCR reaction was performed in a C1000 Touch thermal cycler (Bio-Rad) using the following cycling conditions: 95°C for 10 min, followed by 40 cycles of 94°C for 30 s, 55°C for 1 min, 72°C for 1 min, 98°C for 10 min, and a final incubation at 4°C. Droplets were read using a QX200 Droplet Reader (Bio-Rad), and data were analysed using QuantaSoft analysis software version 1.7 (Bio-Rad). Reactions were performed at least in duplicate, and all runs included no-template controls. DNA samples from Mel-2, A375, WM3282, and WM3406 were used to represent homozygous BRAF^wild-type/NRAS^wild-type, homozygous BRAF^V600E, homozygous BRAF^V600K, and homozygous NRAS^Q61K, respectively.

Correlation analyses were performed with the Spearman rank correlation coefficient.

The DNA samples used for assay validation were obtained from the human melanoma cell lines Mel-2, A375, WM3282, and WM3406. To validate the limits of detection, DNA from the A375 (BRAF^V600E: homozygous), WM3282 (BRAF^V600K: homozygous), and WM3406 (NRAS^Q61K: homozygous) lines was serially diluted into DNA from Mel-2 (BRAF^wild-type, NRAS^wild-type) cells, and the alleles were quantified using ddPCR. The mutated DNA was diluted within a total of 4 ng of DNA template because at least 4 ng of cell-free DNA (> 1 ng/µl) was obtained from patients and control samples for each ddPCR plate well. BRAF^V600E was detected at concentrations as low as approximately 1.6 copies/well in 4 ng of DNA template, and the assay was linear over the entire tested range (R²= 0.995, p < 0.01; Fig. 1a). Similarly, as low as 1 copy/well of BRAF^V600K (R²= 0.971, p < 0.01; Fig. 1b) and NRAS^Q61K (R²= 0.995, p < 0.01; Fig. 1c) was detected in 4 ng of DNA template. Wild-type and mutant alleles were not detected in blank samples without DNA.

Next, cell-free DNA was extracted from the peripheral blood of 5 patients with melanoma who were treated with nivolumab, an anti-PD-1 antibody (Table I). ddPCR was used to quantify ctDNA, which was detected in all 5 patients during their follow-up (Table II). The copy number of BRAF^V600E ctDNA was 0–12,750 copies/ml, that of BRAF^V600K ctDNA was 0–146 copies/ml, and that of NRAS^Q61K ctDNA was 0–348 copies/ml. In addition, 28 healthy control subjects were examined. A BRAF^V600E signal/well (9 copies/ml) was detected in 2 subjects, a BRAF^V600K signal/well (15 copies/ml) in 1 subject, and a NRAS^Q61K signal/well (7.5 copies/ml) in 2 subjects. However, the signals were not reproducible in healthy control subjects when the assay was repeated.

The levels of ctDNA and another marker, LDH (normal range, < 230 IU/l), were assessed in patients during the treatment course.

Patient 1 (BRAF^V600E mutation in the primary lesion) presented with multiple metastases, and BRAF^V600E ctDNA was detected in plasma DNA on day −22 (180 copies/ml, Fig. 2a). Nivolumab administration (day 0) resulted in a decrease in the level of BRAF^V600E ctDNA to undetectable levels on day 21 (Fig. 2a and b). Similarly, LDH levels decreased to the normal range on day 21. A CT scan revealed a significant reduction in tumour size, and tumour response was classified as PR in the RECIST assessment on days 58, 105, and 185 (Fig. 2c). Although the level of BRAF^V600E ctDNA was low until day 213, new lesions were detected in the bladder on day 240, and tumour response was categorized as PD. Furthermore, new lesions appeared in the brain, and the BRAF^V600E ctDNA level was increased on day 268 (105 copies/ml), although the LDH level remained within the normal range. The LDH level exceeded the upper limit on day 351. Changes in ctDNA levels preceded those in LDH. Treatment was changed to vemurafenib on day 352. ctDNA and LDH levels decreased significantly at 1 week after commencing vemurafenib treatment. The patient died on day 427.

Patient 2 (BRAF^V600E mutation in the primary lesion) developed PD with multiple metastases, and BRAF^V600E ctDNA was detected in plasma DNA on day −7 (480 copies/ml, Fig. 3a). The level of BRAF^V600E ctDNA was significantly decreased on day 19 because of nivolumab administration (39 copies/ml). Although the ctDNA level remained low until day 82, a CT scan revealed multiple new lesions in the brain on day 96, which represented PD in the RECIST assessment. The ctDNA level was substantially increased after day 103. Treatment was changed to ipilimumab on day 147.

Patient 3 (BRAF^V600K mutation in the primary lesion) presented with multiple lung metastases. BRAF^V600K ctDNA was detected in plasma DNA at the time of nivolumab initiation (day 0; 146 copies/ml, Fig. 3b). The level of BRAF^V600K ctDNA started to decrease on day 12, decreasing further thereafter and remaining low, with undetectable levels between days 82 and 229. A CT scan showed SD in the RECIST assessment on days 54 and 117. Although the lung metastases were stable, a new lesion appeared in the spleen on day 222, and the clinical evaluation was PD. The patient underwent resection of the spleen metastasis on day 281. Her condition was clinically stable, and the BRAF^V600K ctDNA remained undetectable or low before and after the resection. The LDH level remained above the upper limits of the normal range during monitoring. The ctDNA levels seemed to be consistent with the patient’s clinical condition.

Patient 4 (BRAF^V600E mutation in the primary lesion) developed PD with multiple metastases, and BRAF^V600E ctDNA was detected in plasma DNA on day −12 (1,170 copies/ml, Fig. 3c). Despite nivolumab administration, pleural effusion and ascites progressed and the patient deteriorated rapidly. The BRAF^V600E ctDNA level did not decrease in response to nivolumab. BRAF^V600E ctDNA increased to 10,140 copies/ml on day 30. Although the LDH level increased, it remained within the normal range until day 30. A CT scan showed PD on day 48. The patient died on day 84.

Patient 5 (NRAS^Q61K mutation in the primary lesion) presented with multiple metastases and NRAS^Q61K ctDNA was not detected in plasma DNA at the time of nivolumab initiation (day 0, Fig. 4a and b). The patient’s clinical condition deteriorated despite nivolumab administration. NRAS^Q61K ctDNA and LDH levels increased to 48 copies/ml and 371 IU/l on day 27, and increased further according to the patient’s condition. A CT scan showed PD on day 70 (Fig. 4c). The patient died on day 122.

Monitoring the response to treatment is essential to determine the benefit of therapies and avoid the prolonged use of ineffective and potentially toxic treatments. There is an unmet need to identify biomarkers for measuring treatment response in melanoma. Here, we used ctDNA with BRAF and NRAS mutations in the peripheral blood to monitor the response to anti-PD-1 immunotherapy in 5 patients with advanced melanoma. ctDNA levels were approximately correlated with the response to treatment with anti-PD-1 antibody, suggesting the value of ctDNA as a blood-based biomarker.

The clinical management of patients undergoing immune checkpoint blockade therapy can be challenging. Radiographic changes on conventional CT scans can be misleading because the tumour may show apparent enlargement or new lesions may appear before later regressing. Furthermore, prolonged disease stabilization may contribute to overall survival benefits. However, many cancers such as melanoma have no validated tumour marker. Genetic alterations in human cancers, such as point mutations, can be useful as personalized biomarkers because they distinguish ctDNA from normal DNA. ddPCR has high sensitivity and specificity for the detection of a small amount of DNA. In the present study, ddPCR detected 1–2 copies of mutant BRAF and NRAS DNAs in 4 ng DNA. However, ddPCR also detected mutated DNAs in a few healthy subjects, and the occurrence of false-positive reactions associated with polymerase-induced errors needs to be considered (19). However, the detection of ctDNA in healthy subjects is not usually reproducible because the false-amplification rate is low. In patients treated with BRAF inhibitors, low levels of BRAF ctDNA are associated with a high overall response rate and long progression-free survival (20, 21). As ddPCR capable machines are increasingly available, this approach can be used to monitor tumour burden in melanoma patients in a minimally-invasive manner.

Although molecular-targeted agents, such as BRAF and MEK inhibitors, exert a rapid effect on melanoma, immunotherapy is occasionally associated with a latency of its anti-tumour effect. The molecular-targeted agents attack melanoma cells directly, and ctDNA levels decrease within several days after the administration in cases of favourable response (4, 20). By contrast, in the present study, the decline of ctDNA required 2–4 weeks after the first administration of nivolumab in patients with controlled disease. Cellular tumour activity is inhibited by the immune-mediated effects of the immune checkpoint blockade, and tumour destruction may take a longer time. In addition, the ctDNA levels increased within 2–4 weeks in patients in which the disease was not controlled. An increasing ctDNA level indicates persistent tumour activity and hence worse prognosis compared with patients who have undetectable or falling ctDNA levels. Changes in ctDNA levels in melanoma patients treated with immune checkpoint blockade were recently reported by other groups (22, 23). Consistent with the present results, the therapeutic effect of the immune checkpoint blockade was reflected in the ctDNA levels several weeks after the initiation of treatment.

Secondary resistance occasionally develops during immunotherapy, and it is difficult to determine whether immune checkpoint blockade should be continued. Increased ctDNA may reflect a state of secondary resistance to the treatment in which tumour cells have evaded the immune checkpoint blockade. The levels of ctDNA increased in Patient 1 and Patient 2 despite low and undetectable levels immediately after nivolumab initiation. The rebound in ctDNA levels correlated with treatment failure. On the other hand, ctDNA levels remained low or undetectable in Patient 3 despite the definition of PD on day 222 because of spleen metastasis, and her clinical condition remained stable for more than 6 months after splenectomy, suggesting that nivolumab had an effect on tumour activity. With frequent monitoring, ctDNA may be detected simultaneously with or before radiological detection of progression. Although CT scans are usually performed every 2–3 months, an increase in ctDNA levels could potentially serve as a trigger for early evaluation with imaging modalities, leading to a timely switch to more appropriate therapeutic interventions.

Measurement of LDH levels has been incorporated into the management of melanoma patients, and an elevated level is associated with high disease burden and decreased survival. Recently, LDH was reported as a useful marker to predict early response to nivolumab in patients with melanoma (24). However, LDH is neither sensitive nor specific, and it can be slow in reflecting changes in disease status (23). The ctDNA level, as measured by ddPCR, was a better indicator of treatment response and the emergence of treatment resistance than LDH levels in the present study.

Longitudinal analysis of ctDNA during treatment demonstrated that ctDNA levels detect treatment response and the emergence of resistance and may allow for an alternative treatment to be introduced before a major decline in health parameters. Taking a “wait-and-see” approach may cost a patient valuable time during which suitable alternative interventions could be tried. A limitation of the present study was the small number of patients. Further prospective analysis is required to assess the value of ctDNA as a biomarker of clinical outcome in patients receiving immune checkpoint blockade therapy.

The authors would like to thank Ms. Aya Uchiyama for technical assistance.

The authors declare no conflict of interest.