Quality control of BRAF mutation testing methods used in routine practice is crucial for optimal treatment selection. In this prospective study, we assessed the impact of patient/sample characteristics on BRAF mutation testing results in patients with melanoma, during clinical practice. Data were collected on routine testing practices and documented mutation status in patients with melanoma stages IIIB, IIIC, or IV across 28 diagnostic pathology centers in Germany. Patient/sample data collected included: patient age, location of primary melanoma and metastases, origin of sample, melanoma subtype, and quality of tissue. Statistical influence of patient/sample characteristics on BRAF mutation rate was assessed using multiple logistic regression analyses and statistical models developed to predict the probability of BRAF mutations for individual patient cohorts. Data/samples from 642 patients with melanoma were analyzed. BRAF mutations were documented in 241/642 patients (37.5%). The primary statistical model to predict BRAF mutation rates included: age (continuous), origin of sample, method of mutation analysis, and quality of tissue. Analyses of post hoc collected data identified major deviations between documented mutation rates included in this study vs. routinely recorded mutation rates for three centers. When samples from these centers were excluded, the influence of testing method was no longer statistically significant. The final model included patient age, origin of sample (including metastasis location), and quality of tissue. Once validated in an independent population, this type of model could allow pathology centers to compare the performance of their testing methods with what would be expected based on patient, tumor, and sample characteristics.
Keywords: BRAF mutation; Melanoma; Multiple logistic regression; Mutational analysis; Quality control.