The search for a melanoma-tailored chemotherapy in the new era of personalized therapy: a phase II study of chemo-modulating temozolomide followed by fotemustine and a cooperative study of GOIM (Gruppo Oncologico Italia Meridionale)

Malignant melanoma, although far less prevalent than non-melanoma skin cancers, is the major cause of death from cutaneous neoplasms. MM remains a cancer with a poor prognosis and a chemoresistance profile. However, since 2011, an improvement in overall survival has been obtained thanks to major advances in understanding the driver molecular alterations and the immunogenic potentiality of this unique cancer [1]. The selective inhibitors vemurafenib and dabrafenib, alone or in combination with MEK inhibitors, have achieved a response rate of approximately 50–70%, resulting in improved progression-free survival (PFS) and overall survival (OS) as shown in Phase III studies of patients harbouring BRAF mutations [2, 3]. Nevertheless, a high rate of G3-G4 toxic events ranging from 48 to 63% has also been reported with approximately 15% of patients discontinuing treatment due to side effects. In addition, the majority of patients progressed after approximately 12 months because of the occurrence of numerous mechanisms of resistance to anti-BRAF/MEK drugs [2, 3].

In the immune-therapy field, the immunomodulating antibodies that target the checkpoints CTLA-4 (ipilimumab) and PD1 (nivolumab and pembrolizumab) alone or in combination showed survival benefits as both first and second line therapies. The response rate and the PFS ranged from 15% and 2 months, respectively, for ipilimumab [4] to approximately 40% and 6 months, respectively, for antiPD1. The combination of these drugs resulted in a significant increase in the response rate to 60% with a PFS of approximately 12 months, but its toxicity profile was often unacceptable with G3-G4 side effects reported for over 50% of patients and with therapy interruption in approximately 40% of them [5, 6].

Parallel to the spread of its use, for immunotherapy, many escape mechanisms have been reported so that only a few patients are long-term survivors [7, 8].

Therefore, a considerable number of MM patients receive standard chemotherapy mainly as a subsequent line of therapy. The need to define novel therapeutic strategies that overcome the chemotherapy resistance of MM is still relevant today and represents one of the main challenges in the treatment of advanced disease.

Active chemotherapies in MM include alkylating agents such as dacarbazine (DTIC), temozolomide (TMZ) and fotemustine (FM). DTIC gives an overall response rate of only 10–15% with a complete response in less than 5% of patients and a survival of 7–8 months [9]. Similar overall response rates were achieved with both TMZ and FM. The first drug has a high oral bioavailability with an extensive tissue distribution [10], and the latter has good penetration through the blood-brain barrier but relevant myelotoxic side effects [11].

The activity of alkylating agents depends on their capacity to form alkyl adducts that are made by a chloroethyl group being added to the DNA nucleotide guanine in the case of FM. This action results in DNA interstrand cross-links, which in turn trigger the apoptotic cascade. However, the antineoplastic activity of these agents is limited by cellular resistance principally induced by the DNA repair enzyme O(6)-methylguanine DNA-methyltransferase (MGMT), which removes the chloroethyl group from the DNA strands before the crosslink is established [12].

The depletion of MGMT can reverse resistance to alkylating agents and seems to be induced by continuous drug administration as documented in laboratory research and clinical trials [12–15].

To date, the use of TMZ as a chemo-modulating agent has never been tested in an MM patient population. We evaluated this hypothesis in a feasibility study that included two cohorts of patients treated with two schedules of TMZ (100 mg/m² over 2 days) in combination with FM (100 mg/m² on the second day 4 h after TMZ) in order to identify the optimal doses and timing of administration according to an acceptable safety profile and a strong antitumour activity [16]. We found that this chemotherapy regimen was better tolerated in terms of myelotoxicity when it was administered on a schedule of day 1–21 rather than on days 1 and 8 every 21 days [16, 17].

Thus, we planned a new multicentre phase II trial to verify the effectiveness of this treatment schedule in a larger population of patients. Moreover, we attempted to build a translational study by evaluating a posteriori some biological parameters implicated in drug resistance in order to unearth candidate novel biomarkers that are suitable as predictive and prognostic tools to help us identify responsive patients and optimize the use of these “old” drugs.

It has been frequently asked whether there is a role for chemotherapy in MM considering the numerous drugs available today. The response rates to combination target or immune-therapy with antiBRAF/antiMEK and antiCTLA4/antiPD1 range from 58 to 69%, and the disease control rate is 75% of patients receiving both of these therapies. However, most patients progress after approximately 12 months of treatment, and only a few of them achieve long-term control of their disease. Moreover, the toxicity profile of these new drugs is often unacceptable with G3-G4 side effects reported for over 50% of patients, causing many to discontinue the drugs. Finally, these drugs are often not indicated for patients with various types of comorbidities such as autoimmune, ocular and cardiac diseases. Therefore, the need for other therapeutic options is still very important.

We conducted the first large clinical study for MM, aiming to explore the effectiveness of sequential non-therapeutic, chemo-modulating low doses of TMZ after a full dose of FM. Currently, few data are available, and no dosing or schedules have been established. Additionally, the optimal interval between the administration of the two drugs is not yet clear. A depletion in MGMT can be gained in melanoma cells when TMZ is administered at a low dose of 100–200 mg/m² consecutively for 2 days. This enzymatic deficiency can amplify the effectiveness of FM when it is given on the second day approximately 4 h after TMZ [12–14]. In MM, two previous studies have tested the combination of TMZ with nitrosureas, namely, FM [17] and lomustine [18]. In both of these trials, TMZ was given at a higher dose than our schedule and with an additive/synergistic intent in combination with the full dose of nitrosureas. As a consequence, an unacceptable toxicity with a higher rate of myelotoxicity was reported in both studies. In particular, Tas et al. [17] reported a dose reduction in 45% of the patients, a dose delay in 32.5%, with a toxicity related discontinuation of 27.5%, a response rate of 35% and a low median survival of only 6.7 months.

We used a regimen previously verified in our pilot study [16]. In the present study, in a large cohort of 69 MM patients, we confirmed a response rate of 30.3% and an overall clinical benefit of 50.5%. The median PFS was 6 months, and the median OS was 10 months. Notably, our patient population included 74% patients in the M1c stage, of whom 15% had brain metastases. This means that this population had a very poor prognosis.

When we compared patients who obtained a clinical benefit (SD + PR + CR) versus patients with progressive disease, we found a median PFS of 7 versus 3 months and a median OS of 14 versus 6 months. These data mean that a huge effort should be made to tailor these drugs to selected patients through the identification of biomarkers.

In our previous proteomic study carried out in 20 patients of this same population, we identified some peptides that were significantly upregulated in responder patients and associated with proteins involved in the control of redox cellular homeostasis, such as NQO1, and in the regulation of apoptosis, such as RIN1 [19].

In this translational effort, we also explored if the effectiveness of our schedule was predicted by the level of MGMT methylation. We found low levels (6–13%) of pretreatment MGMT methylation, which were not related to the clinical response. Our findings are in accordance with previous data showing that in MM no association exists between the clinical response to chemotherapy and basal levels of MGMT [20–22]. Otherwise, it has been reported that low MGMT nuclear expression, evaluated by immunohistochemistry, is associated with better outcomes only in patients with BRAF mutations treated with a cisplatin, vinblastine and temozolomide regimen as the first-line therapy [23], and in glioblastoma, MGMT methylation greater than 35% has been described as an independent prognostic factor associated with better outcomes [24, 25].

Notably, MGMT is not an exclusive player involved in melanoma cell death induced by alkylating drugs [12, 25]. The inherent deficiency of the downstream apoptotic pathway might be a key resistance mechanism, and this might be due to several sources such as mismatch repair protein inactivation and alterations in DNA damage repair pathways.

Thus, we analysed the expression of genes involved in the base-excision repair (BER) pathway because of their emerging importance in enhancing the cytotoxicity of DNA damaging agents (e.g., alkylating agents). We were able to collect FFPE samples for only 14 patients before treatment. Notwithstanding the small sample size, gene expression of APE1, XRCC1 and PARP1 was measured to verify a trend that could explain the response to treatment. APE1 is involved in a key step of BER and has an almost unique role in the processing of apurinic/apyrimidinic sites [26]. We observed that the basal mean level of APE1 gene expression was elevated in patients who did not respond to treatment versus those who responded to chemotherapy. Abbots et al. [27] reported that APE1 inhibition is efficient in PTEN-deficient melanoma cell lines, and our results encouraged us to further investigate the role of this enzyme in MM. In a similar way, we observed the upregulation of protein 1 of the PARP family in patients who progressed after a few cycles of TMZ/FE treatment. Moreover, patients with PARP1 downregulation showed a longer median OS rate.

XRCC1 is a scaffold protein with no enzymatic activity that interacts with several components of the BER pathway. Its deficiency is responsible for mutations and a high rate of sister chromatid exchange, which leads to genomic instability. It has been reported that such a deficiency results in chemo-sensitivity [28]. Abdel-Fatah et al. [29] reported that a deficiency in XRCC1 in ovarian cancer is associated with a clinical response to cisplatin treatment. In accordance with this study, we observed a slightly elevated mean expression level in non-responding patients, although the survival analysis showed that patients with upregulated expression had a longer median OS rate. This result seems to confirm that of another study reporting that wild-type XRCC1 cell lines are more sensitive to TMZ and, more interestingly, that effective PARP inhibition requires a functional XRCC1 protein [30].

Although increased expression of BER genes we observed in the not-responding patient group was not significant, our preliminary results encourage verification of the role of players in the BER pathway in melanoma treatment both as predictive biomarkers, such as XRCC1, and as molecular targets (PARP1 or APE1) in order to enhance current therapeutic settings.

In this large phase II trial, we demonstrated that the combination of two “old” alkylating agents effectively works in terms of both overall response and survival with an acceptable toxicity profile. In view of the increasing range of therapeutic options now available, an emerging challenge for clinicians is to establish a useful algorithm of sequential treatment for MM patients. Chemotherapy can still play a role, mainly in BRAF wild-type patients who progress on immune therapy or for whom immunotherapy is contraindicated. Additionally, in patients with mutated BRAF, chemotherapy can be utilized mostly in cases of fast progression during targeted therapy. For this purpose, we must endeavour to shed light on the mechanisms underlying drug responses and resistance as well as to outline useful biomarkers that could help us to tailor the optimal agents to the appropriate subjects at the right time.