Predicting Outcomes in Metastatic Melanoma

Our 1983 study, published in the first volume of the Journal of Clinical Oncology (JCO), was the fourth in a series of multifactorial analyses of melanoma patients treated at the University of Alabama at Birmingham (Birmingham, AL) and was the first paper published on melanoma in JCO. These were the first studies using the Cox multifactorial analysis for analyzing prognostic factors in melanoma, including this analysis of distant metastases (designated as stage III in the 1980s and stage IV today). These publications were not simply the result of adopting a new statistical methodology and providing statistical services to clinical investigators. A major component of our success can be ascribed to the synergistic research collaboration that resulted from blending together our different training and perspectives. Thus, we spent considerable time teaching each other about melanoma and about iterative statistical approaches to address a series of hypotheses that were mutually agreed on. Although the number of patients with distant metastases was relatively small in this 1983 publication, they were a representative population from a single institution. Remarkably, the independent prognostic factors we identified for stage IV melanoma almost 25 years ago remain valid today. We attribute this to the quality of the data, the use of sophisticated statistical methodology, and the meaningful collaborations between clinical oncologists and biostatisticians. In addition, these survival results probably reflect the true natural history of distant metastatic melanoma without the influence of systemic therapy on survival rates, given that there were few systemic treatment options during the 1970s when these patient results were accrued. Thus, the number and site of distant metastases and the remission duration still are the primary independent factors for stage IV melanoma; these results have been validated by subsequent published studies. The introduction of the Cox proportional hazards regression model represents the most important methodologic development for multivariate analysis of survival data during the last three decades. In melanoma research, a large number of clinical factors (eg, age, sex, lesion site, performance status) and pathologic factors of the primary tumor (eg, tumor thickness, tumor ulceration, mitotic rate, level of invasion, growth pattern) and the metastatic tumor (eg, number, site or sites, size). Although these factors related to disease recurrence and patient survival were all studied extensively before the 1980s, the results were variable depending on such factors as sample size, case mix, and statistical methods. With an application of the Cox regression model, almost all major multivariate prognostic factor studies have identified a remarkably consistent set of independent prognostic factors for melanoma patients treated worldwide. In addition, several useful predictive models based on the Cox model for predicting individual patient survival and disease recurrence in melanoma were also developed from several large melanoma databases. These advances, in turn, have facilitated a fundamental revamping in the staging of melanoma and the criteria for interpreting results of prospective clinical trials in melanoma using the dominant prognostic factors identified by Cox regression analyses. Since the year 2000, we have collaborated with melanoma clinical investigators worldwide to create a unique melanoma staging and prognosis database under the auspices of the American Joint Committee on Cancer and the International Union Against Cancer. The first version of this Melanoma Staging Database incorporated the clinical and pathologic results of more than 17,600 prospectively observed melanoma patients treated on three continents. The results using the Cox multifactorial methodology led to a major revision of the melanoma staging criteria and stage grouping. These results were first published in JCO in 2001. Since 2007, an updated American Joint Committee on Cancer Melanoma Staging Database has been created that contains pathology and treatment outcome data on more than 50,000 prospectively observed melanoma patients treated in the United States, Australia, and Europe. Results of the data analysis are still preliminary. The prognostic factors described above are still the dominant clinical and pathologic features of melanoma for stage I, II, and III melanoma, with the exception that tumor mitotic rate and patient age are highly predictive and independent predictors of outcome. The stage IV data analysis is still pending. A mathematical predictive model has been developed and validated that will provide an electronic predictive tool that enables the integration of multiple and continuous variables to predict in an individual patient the risk of regional and distant metastases, and the actuarial melanoma-specific survival rates at 5 and 10 years. Although the Cox model has demonstrated its flexibility and usefulness in modeling survival data in melanoma, it also has inherent limitations because of its proportional hazards assumption and its inability to generate a hazard function. We are now using a new parametric statistical modeling that can integrate all independent predictive factors and thereby calculate an individual patient’s predicted outcome at the onset of his or her disease stage and at any time point thereafter. We have demonstrated that the parametric model can serve as a powerful alternative or JOURNAL OF CLINICAL ONCOLOGY C E L E B R A T I N G 2 5 Y E A R S O F J C O VOLUME 26 NUMBER 2 JANUARY 1