Changing treatment landscape Novel therapies including immune checkpoint inhibitors (ICIs) and BRAF/MEK targeted drugs have fundamentally changed melanoma management. However, only 30-40% of patients respond to ICIs, creating urgent need for imaging biomarkers that can predict response and stratify patients before treatment.
Role of 18F-FDG PET/CT Positron emission tomography with fluorodeoxyglucose (18F-FDG PET/CT) exploits cancer cells' heightened glucose uptake to detect tumors metabolically. It has an established role in staging advanced melanoma and evaluating therapy response, and its derived parameters - SUVmax, metabolic tumor volume (MTV), and total lesion glycolysis (TLG) - carry prognostic significance.
Emerging radiomics Radiomics applies AI-derived quantitative texture analysis to medical images, extracting features invisible to the human eye. This review examines both conventional PET/CT metrics and the emerging radiomics field in the context of melanoma management.
Stage-dependent utility PET/CT has limited sensitivity for detecting nodal metastases at early stages (AJCC I-II) compared to lymphoscintigraphy and sentinel node biopsy, which remain the standard for these patients. Its utility is much higher for advanced-stage (III-IV) disease, where sensitivity reaches 68-87% and specificity 92-98%.
Impact on clinical management In a retrospective study of 59 patients, 18F-FDG PET/CT detected greater disease extent than CT alone in 55% of subjects and changed the treatment plan in 49% of cases. In another prospective study, PET/CT upstaged 16% of patients with in-transit or satellite metastases and detected new distant metastases during follow-up in 40% within 6 months.
SUVmax as a prognostic parameter A prospective study of 80 stage IIIB melanoma patients found that SUVmax measured on nodal metastases before surgery significantly predicted outcomes. Patients with low SUVmax had a 5-year disease-free survival of 41% versus 24% for those with high SUVmax, confirming PET metrics as useful pre-therapeutic stratification tools.
Early metabolic response detection In BRAF-mutated melanoma patients treated with vemurafenib, 18F-FDG PET/CT performed at baseline and just 15 days after therapy initiation detected metabolic response before anatomic tumor shrinkage was evident on CT. A 25% decrease in SUVmax (EORTC criteria) correlated with a trend toward longer progression-free survival.
Intra-individual response heterogeneity A study of 23 patients receiving dabrafenib found homogeneous metabolic response in 74% and heterogeneous response in 26%. Critically, patients with heterogeneous metabolic response had shorter time to progression (3.0 versus 7.4 months), identifying heterogeneity as an adverse prognostic feature detectable by PET.
Volumetric parameters and resistance In 57 BRAF-mutated melanoma patients on BRAF/MEK inhibitor therapy, high baseline metabolic tumor volume (MTV greater than 56 cc) and involvement of more than 2 metastatic organ sites were independently associated with shorter progression-free and overall survival, even after multivariate adjustment.
Pseudoprogression challenge Immunotherapy can cause initial tumor enlargement due to immune cell infiltration before true response occurs - a phenomenon called pseudoprogression. Standard response criteria (RECIST) may wrongly classify this as treatment failure. Immune-modified RECIST (iRECIST) and PET-based criteria were developed to distinguish pseudoprogression from true progression.
Multiple PET criteria for ICIs Several PET-specific response criteria have been developed for immunotherapy monitoring including EORTC, PERCIST, and imPERCIST. The imPERCIST criteria permit continued therapy even with initial progression if subsequent imaging confirms benefit, addressing the unique kinetics of immune response.
PET-derived parameters predict ICI outcomes Multiple studies show that baseline MTV and TLG independently predict overall survival in melanoma patients treated with ipilimumab, nivolumab, or combination ICIs. High baseline MTV is consistently associated with poor outcomes, even in patients who achieve initial metabolic response.
What radiomics offers Radiomics extracts hundreds to thousands of quantitative features from medical images - including first-order statistics, texture, and shape parameters - that are invisible to radiologists. Applied to 18F-FDG PET/CT, radiomics can potentially capture the biological heterogeneity of melanoma in a non-invasive way.
Predicting mutation status Radiomic features extracted from PET/CT images have been explored to predict BRAF mutation status, which determines eligibility for targeted therapy. If validated, such tools could spare patients from invasive molecular testing, particularly when tissue biopsy is difficult or heterogeneous tumors are present.
Response prediction studies Early radiomics studies in melanoma show promise for predicting immunotherapy response before treatment begins. Textural heterogeneity features from baseline PET images correlate with subsequent response to anti-PD-1 and anti-CTLA-4 therapies, potentially enabling better patient selection.
Biomarker-guided treatment selection The combination of PET-derived metabolic parameters (SUVmax, MTV, TLG) with emerging radiomics signatures offers a non-invasive path toward individualized treatment selection in melanoma. This is particularly important because only a minority of patients benefit from expensive immunotherapies with significant side effect profiles.
Monitoring and switching strategies Serial PET/CT enables timely detection of both treatment response and acquired resistance. Identifying BRAF-inhibitor resistance early through heterogeneous metabolic response patterns allows switching therapy - for example from targeted to immunotherapy - before significant clinical deterioration.
Brain metastases limitation A key limitation of 18F-FDG PET/CT is its poor sensitivity for brain metastases due to high background brain glucose metabolism. MRI remains the gold standard for intracranial staging. Radiomics applied to MRI rather than PET is therefore being explored separately for characterizing melanoma brain metastases.
Need for standardized criteria Multiple competing PET response criteria (EORTC, PERCIST, imPERCIST) exist for melanoma immunotherapy monitoring without consensus on which is optimal. Prospective head-to-head comparisons are needed to standardize clinical practice and allow cross-study comparison of radiomics models.
Radiomics validation challenges Most radiomics studies in melanoma are retrospective and use small, single-center cohorts. Prospective multi-center validation with standardized image acquisition protocols is essential before radiomics signatures can enter clinical practice as reliable biomarkers.
Future directions Integration of PET/CT radiomics with liquid biopsy data, genomic profiles, and clinical variables in multi-modal AI models represents the logical next step. Such models could enable truly individualized treatment decisions, predicting not just who will respond but who will develop immune-related adverse events.