Customize your JAMA Network experience by selecting one or more topics from the list below.
The current issue of JAMA Dermatology contains 3 articles that attempt to describe some aspect of disease kinetics of selected subsets of patients with melanoma.
The first of these studies, from von Schuckmann and colleagues1 in Australia, attempts to describe factors predictive of early recurrence (within 2 years) among patients who presented for treatment of clinically localized T1b to T4b melanoma in Queensland. Patients were recruited prospectively from a number of hospital and private clinic sources. Of 1245 patients identified and invited to participate, 700 (56%) were included in the analysis. Most of the follow-up was obtained as a patient-reported binary recurrence status (yes/no) every 6 months. In the event of patients lost to follow-up, local hospital records and Queensland Cancer Registry data were reviewed. All recurrences were verified insofar as possible by medical record review. Factors identified predicting recurrence within 2 years were not surprising and included the presence of ulceration, an increased mitotic index, increasing T stage, and head and neck location. Descriptive statistics were provided comparing outcomes by the seventh and eighth editions of the AJCC Staging Manual, and describing the initial sites of recurrence.
The biggest challenge in interpreting the significance of this study is the heterogeneity of the patient data set, a data set that included 442 patients with clinical T1b to T4b who did not undergo sentinel lymph node biopsy (SLNB), 213 patients with pathologic T1b to T4b who had a negative SLNB result, and 38 patients with at least stage IIIa after a positive SLNB result (the SLNB numbers do not quite add up to 700 in Table 1). This heterogeneity recalls that of the AJCC7 data set, where pathologic staging of clinically localized melanomas was not required. The AJCC8 has addressed this shortcoming by including only patients who underwent SLNB,2 thus describing more homogeneous T-stage subsets. Excluding the patients with T1a disease, the AJCC8 data set depicted describes the dynamic kinetics of outcome in 10 466 patients with pathologic T1b to T4b melanomas. As expected, more advanced tumors generally recur more frequently and earlier. Importantly, the AJCC8 data set demonstrated that most events in this cohort of patients are expected to occur after the 2-year threshold described in the study by von Schuckmann et al,1 with continued attrition out to at least 10 years. Thus, the second concern: the 2-year interval, although entirely practical for a prospective study such as this, may not be entirely reflective of the true biology of this heterogeneous cohort.
The second study, from Vallet and colleagues in Paris,3 focused on the question of whether, in a cohort of patients with established distant recurrence after treatment of a known antecedent primary melanoma, the time from initial primary excision to the first distant recurrence was associated with progression-free (PFS) and overall (OS) survival (although not explicitly stated, these are assumed to be postrecurrence PFS and OS). The investigators queried MelBase, a multicenter collaborative French database of patients with unresectable stage III and stage IV melanoma. They created a contemporary study cohort of 638 patients with stage IV melanoma who met their inclusion criteria, accrued between March 2013 and September 2017. Only patients with complete information on all required dates receiving first-line systemic monotherapy and/or in a clinical trial (immune monotherapy, 43%; targeted monotherapy/clinical trial, 28%; clinical trials [otherwise unspecified], 21%; and single-agent chemotherapy/clinical trial, 10%) were included. Unless included in the clinical trial group, patients receiving nonprotocol combination ipilimumab/nivolumab, patients treated with nonprotocol BRAF/MEK inhibitor combinations, patients receiving multiagent chemotherapy, and patients treated with up-front surgical resection of stage IV disease were apparently excluded. The data set appears to contain some minor irregularities: 13 patients were staged as M0, 15 patients had AJCC stage III, and 20 patients had an antecedent AJCC stage of 0 at primary excision prior to the diagnosis of stage IV disease. Furthermore, 55 patients had either mucosal or “other” anatomic sites (noncutaneous?), melanoma subsets with biologies potentially quite distinct from cutaneous melanoma. The data set might have been cleaner had these patients been excluded, with the analysis confined to patients with stage IV cutaneous melanoma only.
Notwithstanding these minor concerns, the authors observed that the interval from initial treatment to the first distant metastasis, whether analyzed as a discrete or continuous variable, was not associated with (postrecurrence) OS or PFS. Once again, in the absence of a more robust statistical evaluation of other factors recorded in Table 1 that might have been prognostic for OS/PFS, such as metastasis substage, the presence of brain metastases, the number of metastases/metastatic sites, Eastern Cooperative Oncology Group status, treatment type, and lactate dehydrogenase levels, it is difficult to put this single observation into its proper context. Although the authors allude briefly to Cox models showing no interaction between time to first distant recurrence and Breslow thickness or stage at initial treatment, they appear to have missed the opportunity to fully explore and display a more in-depth multivariable analysis of their data.
Older historical references cited by the authors came from a time before the advent of meaningful systemic therapy, when the trajectory of a discrete minority of highly selected patients with stage IV melanoma could be affected only by complete surgical resection. Parameters routinely cited as associated with improved outcome, and occasional long-term survival, included a low number of metastatic sites, and relatively crude clinical measures of what were thought to reflect an individual tumor’s inherent biologic metronome, measures such as disease-free interval prior to diagnosis of stage IV disease or tumor doubling time. Most authors have concluded that, at least when selecting patients for potentially curative metastatectomy, the more stringent the patient selection criteria were, the more likely it would be that resection would alter the course of a patient’s disease. Interestingly however, in the recent study by Faries et al4 of the end results of the MMAIT randomized trial of metastatectomy with or without immune therapy for stage IV melanoma, with fairly broad inclusion criteria (resection of up to 5 metastatic sites), not only did the authors report a 5-year OS rate of around 40% after resection, in multivariable analysis, the time from primary diagnosis to randomization was not prognostic. This would seem to be consistent with the observations of Vallet et al.3
There is another potential explanation for the observations of Vallet et al3 in this very contemporary cohort of patients with stage IV disease, one that most melanoma oncologists would prefer to embrace. In sharp contrast to bygone times, we now have therapy that can actually alter the kinetics of metastatic melanoma. As complete resection of metastatic disease has always been the most powerful prognostic variable among patients with melanoma undergoing metastatectomy with curative intent, now it is quite likely that a patient’s best response to immune and/or targeted therapy will emerge as much more important than the time it took for stage IV melanoma to become apparent. Response to systemic therapy, seen only anecdotally in the chemotherapy era, is now observed with increasing frequency, with clinical benefit defined as ranging from stabilization of previously progressive disease to increasingly frequent and durable complete remission of all known disease. Although there has been some thoughtful interest in exploring the prognostic value of establishing the kinetics of stage IV melanoma as a prognostic factor prior to initiating therapy,5,6 there has been limited uptake of this approach by the melanoma oncology community. Studying the natural history of untreated advanced melanoma is becoming increasingly obsolete.
In the third article, by Calomarde-Rees and colleagues7 from Valencia, the investigators looked at clinical and pathologic clues to distinguish which primary melanomas were more prone to initial lymphatic or to initial hematogenous metastases. This is a retrospective analysis of a large single-institution database, identifying 1177 patients with stage I/II melanoma who underwent wide excision, and SLNB if primary tumors were more than 0.75 mm thick. In addition to appropriate surgical staging data, the authors also had the mutational status of 3 genes in a substantial minority of patients: BRAF (540 [46%]), NRAS (532 [45%]), and TERT promoter (306 [26%]). At a median follow-up of 75 months, 151 patients had experienced an initial recurrence, 43 lymphatic only, 43 systemic only, and 66 combined lymphatic/systemic. Independent predictors of any initial lymphatic recurrence included age older than 55 years, tumor location, increasing tumor thickness, and lymphovascular invasion/satellitosis. Independent predictors of any initial hematogenous recurrence included increasing tumor thickness, presence of a TERT promoter and/or BRAF mutation, and the absence of primary tumor regression. In a second statistical model, in addition to increasing tumor thickness and the absence of regression, there was a particularly adverse association between any initial hematogenous metastasis and tumor harboring both a TERT promoter mutation and mutation in either BRAF or NRAS.
That increasing tumor thickness was associated with increased risk of recurrence, either lymphatic or hematogenous, comes as no surprise. That lymphovascular invasion/satellitosis was associated with initial lymphatic metastasis is also intuitive; satellitosis is already considered a criterion for stage III in both AJCC7 and AJCC8. At the very least, these 2 observations serve as internal controls that validate the credibility of the data set. The associations between older age and tumor location are a bit more cryptic, and may or may not represent true biologic phenomena. Several earlier large studies8-10 have shown that, although older age is independently associated with a lower incidence of a positive SLNB result, it is also independently associated with a higher disease-specific mortality. Similarly, the absence of primary tumor regression as a predictor of hematogenous metastases has long been debated, with strong proponents on either side of the argument. More recent larger series and 1 meta-analysis have concluded that primary tumor regression has either no effect or even a protective effect on disease-specific mortality, the latter being consistent with the observations made in this paper.11,12
The most intriguing finding of this paper is the strong potential association of the combination of both MAPK (either BRAF or NRAS) and TERT promoter mutations with poor survival, as demonstrated in their second prognostic model (HR, 5.7; 95% CI, 2.2-14.5). The authors clearly acknowledge the fatal flaw of this conclusion. Assessment of the mutation status of each gene in their study was performed only in a minority of the study population. We are given no information on which patients were selected for genetic profiling or when it was performed. For example, were the genes tested disproportionately only in patients after they developed metastatic disease, to assist in directing therapy? Were these gene mutations discovered in the context of multigene testing, or were these the only ones examined? To formulate a more robust hypothesis, the authors should confine their analysis to the outcomes of patients who had information on the mutation status of all 3 genes (and any others of interest), done exclusively on the primary tumor (not recurrent tumor), to examine associations between patterns of mutations and subsequent outcomes. Consistent with the findings of this study, although BRAF mutation alone is known to be highly predictive of response to RAF inhibition, it is not generally thought to be prognostic of the time to first distant metastasis.13 The potential interaction of the mutated TERT promoter gene is provocative, especially in the context of a recent article suggesting a role for monitoring BRAF and TERT circulating free DNA as an indicator of response to systemic therapy and outcome,14 but this needs a more rigorous analysis of available data. The role of gene expression profiling in predicting the biologic behavior of melanoma is likely to be much more complex than the mutation of 1 or 2 genes. Numerous investigators have attempted to develop larger gene sets capable of predicting outcome of patients with localized melanoma, independent of or in conjunction with known prognostically relevant clinicopathologic factors.15-18 It remains to be demonstrated whether any of these highly discrepant gene profiles will outperform optimized contemporary multivariable individual patient risk prediction models across the prognostic spectrum of melanoma.
The overall natural history kinetics of locoregional melanoma have been extensively documented.19,20 The AJCC staging system provides evidence that for each successively higher stage, events (recurrence or death) occur earlier and more often, with few late events observed in patients with more advanced disease.
In the new era of effective systemic treatment for melanoma, the kinetics of melanoma progression are no longer confined to the natural history outcome curves we have generated for so many years. They are now highly dependent on response to treatment received. As an example, the RFS curves of patients with stage IIC/III melanoma with and without adjuvant dabrafenib/trametinib demonstrate the contemporary natural history of frequent early recurrences, with few late events in the placebo groups.21 In contrast, the treated groups demonstrate fewer and later recurrences, with an as yet unknown frequency of late events. Effective treatment of systemic disease has forever altered the kinetics of events in advanced melanoma. Hopefully this trend will continue as additional and more effective therapies become available in the years ahead.
Corresponding Author: Daniel G. Coit, MD, Memorial Sloan Kettering Cancer Center, 1275 York Ave, New York, NY 10065 (firstname.lastname@example.org).
Published Online: May 1, 2019. doi:10.1001/jamadermatol.2019.0200
Conflict of Interest Disclosures: None reported.
Coit DG. The Changing Kinetics of Advanced Melanoma. JAMA Dermatol. Published online May 01, 2019155(6):657–659. doi:10.1001/jamadermatol.2019.0200
Create a personal account or sign in to: