At an average administered activity of 198 MBq, no adverse events of greater than grade 1 severity, or laboratory abnormalities attributable to tracer administration were recorded. This is consistent with the absence of toxicity demonstrated in clinical studies involving structurally related benzamides [19, 21]. The whole-body radiation dose was comparable to those calculated for the current standard PET radiotracer, FDG [22, 23]. The bladder had the highest radiation dose, which could be limited by adequate hydration and frequent voiding. Accordingly, there are no safety concerns regarding the use of MEL050.
However, in this small series, the diagnostic performance of MEL050 was substantially inferior to that with FDG, recognizing the potential for a pre-test selection bias with a positive FDG PET/CT being required for study entry. Nevertheless, the tumour to background ratios were also generally lower in our series compared to FDG. A possible explanation for this finding was the demonstration of more rapid metabolism of MEL050 than anticipated based on prior experimental evidence. For example, preclinical studies in C57BL/6J mice suggested high in vivo stability of MEL050 out to at least 2 hours [17]. This was reflected by the low initial bone uptake on serial imaging, with washout similar to other organs, indicating low levels of defluorination, and in metabolite analysis, which showed that 90% of activity from tumour, the eyes and urine, and 70% of activity in plasma, was unchanged at this timepoint. Thus, although tracer stability was high in rodents, MEL050 stability in vivo in humans was relatively low, with a mean 41% degradation in the initial 5 minutes after administration.
Despite this, the biodistribution of MEL050 was suitable for imaging detection of metastatic sites with predominantly renal excretion, and substantially lower brain and bowel activity than is usual on FDG PET scans. Conversely liver, stomach and bone marrow activity remained slightly elevated in comparison to FDG at likely clinical scanning times but was still relatively low. Late visualization of the gallbladder suggests the presence of some hepatobiliary excretion of either MEL050 or of a metabolite. Even though these differing tissue background activities and excretory patterns could impact on the ability of MEL050 PET scans to identify metastases in these locations both positively and negatively, data from this study did not allow a direct evaluation of this possibility with no known brain or gallbladder metastases identified.
The ability of melanin-binding tracers to identify brain metastases that are not apparent on FDG PET/CT has recently been demonstrated using a novel benzamide, [18F]-PFPN, that has structural similarity with MEL050 [24]. In a study involving 21 patients imaged with either PET/CT or PET/MRI, this agent demonstrated significantly higher SUVmax values than on FDG PET, which translated into improved detection of metastatic sites. The lower intensity of uptake and slightly elevated background activity relative to FDG in our series may account for the poor sensitivity of MEL050 PET for detection of metastatic sites relative to FDG PET/CT.
In contrast, the absence of MEL050 accumulation in 5/5 tumours for which pathology demonstrated a lack of pigment, as well as lack of uptake at non-tumoral sites of FDG uptake, suggests a high specificity of MEL050 for the presence of melanin, as demonstrated in animal models [18]. Furthermore, the detection of definite specific ocular localization of MEL050 in 8/10 participants despite substantial partial volume effects related to the minimal thickness of the melanin layer in the retina indicates the capacity of this investigational agent to sensitively detect melanised tissues with PET. Additionally, true positive scans occurred in all 4 tumours in which pathological confirmation of melanin content was available (including 2 tumours with only sparse pigment identified), substantiated the potential of MEL050 to identify tissue melanin. Notably, substantial heterogeneity of uptake was noted within individual patients, suggesting loss of melanin in some tumours rather than a low affinity of MEL050 for melanin.
Melanomas often have variations in pigmentation. In fact, whereas benign pigmented lesions are generally uniform in colour, colour variegation is a classical sign of melanoma, being the third of four key clinical features first outlined by Friedman in 1985 [25]. Lesions with FDG-avidity and lacking uptake of a melanin-targeted tracer have also been recently demonstrated in a larger series using [18F]PFPN [20]. Clinically, heterogeneity of melanin expression in metastatic lesions may limit the potential of melanin-binding agents for theranostic application, which was one of our motivations for developing a range of iodinated benzamides with structural homology with MEL050 [26].
Although amelanosis is thought to occur infrequently in primary melanomas, prompted by the PET findings of this study, we assessed tumour melanin content in a post hoc analysis of 253 patients previously prospectively enrolled in an evaluation of primary melanoma specimens and found an amelanosis rate of 18%. In this cohort, amelanosis in primary melanomas was an independent negative predictor of melanoma-specific and relapse-free survival. This may relate to the delayed diagnosis of amelanotic lesions allowing them to become more advanced and therefore more likely to spread. However, in the cohort with positive sentinel node biopsy, amelanosis in lymph node metastases was a positive prognostic finding. This finding concurs with a recent study of the melanin-targeted agent, [18F]PFPN, in which high SUV levels, suggesting a high melanin content, were associated with a poor prognosis [20]. These apparently contradictory findings suggest complex, disease stage-related relationships between melanin deposition by melanoma cells and the metastatic activity of these cells.
Interestingly, in patients with paired pathological samples, amelanosis was more common in nodal metastases than in primary melanomas, suggesting phenotypic plasticity in metastasizing clones, which is a recognized feature of melanoma [27]. Two plastic phenotypic states have been described in melanoma: a proliferative state driven by micropthalmia-associated transcription factor (MITF) and an invasive state linked to upregulation of the tyrosine protein kinase receptor AXL, which is involved in epithelial-mesenchymal transition (EMT). Accordingly, the tendency to metastasize may not necessarily be linked to the rate of progression after metastasis has occurred.
Although the diagnostic performance of MEL050 was only an exploratory endpoint of the clinical trial and our evaluation of it was limited by the small sample size of our study, the results discouraged our group from further clinical development of this tracer until such a time as the biological significance of loss of pigmentation in some or all sites of identified disease could be identified. The results of pathological sampling of metastatic nodes suggesting a better prognosis for amelanotic than pigmented lesions support recent findings that a higher SUVmax in tumour sites imaged with [18F]PFPN is associated with a superior prognosis [20], but is potentially subject to sampling error and does not necessarily reflect the behaviour of systemic metastases once established.