Predicting circulating biomarker response and its impact on the survival of advanced melanoma patients treated with adjuvant therapy

Abstract

are the melanoma-inhibiting activity (MIA) and the calcium binding protein S100B 3 , but no consensus exists on their prognostic capability. Proper assessment of the predictive capacity of biomarkers longitudinal data should be done in the context of mechanistic computational models linking them with clinical outcome. Biomarker trajectories are usually not linear and show great variability across individuals. Consequently, a non-linear mixed effects (NLME) modelling approach provides a valuable option to handle and model this type of dynamic behaviour. In NLME models, individual profiles are characterized by a common structural model with fixed population parameters and a statistical model with random effects to allow the parameters to vary within the patient population. In this work, longitudinal biomarker data has been described based on semi-mechanistic pharmacokinetic-pharmacodynamic (PKPD) type models and linked to the PFS and OS. Recent efforts have shown that this approach is feasible to identify robust markers that allow the selection of patients that could obtain a therapeutic benefit from the different anticancer treatments and to improve the prediction of their survival 4-6. Therefore, in this study we aim to establish a quantitative treatment-biomarker-survival modelling framework using nonlinear mixed effects PKPD modelling to link the survival of advanced melanoma patients with LDH, MIA and/or S100B protein kinetics following IFN-α b 2 administration. In addition and taking into account the toxicity associated to IFN-α b 2 administration, neutropenic effects were also described mechanistically 7 in the current evaluation providing a highly valuable approach in which to evaluate possible predictors of clinical response while minimizing adverse effects.

Longitudinal and time‐to‐event modeling for prognostic implications of radical surgery in retroperitoneal sarcoma

“…In contrast to previous studies, our modeling framework utilized the entire longitudinal profiles of key biomarkers, rather than crosssectional data, thus enabling a more accurate assessment of the quantitative relationship between each biomarker T A B L E 3 Parameter estimates of the final TTE models and survival. 17 Accordingly, personalized prediction of survival probabilities for different scenarios could be performed using the established parametric TTE models. Besides, a nonlinear mixed effects modeling strategy enabled model-based extrapolation of missing longitudinal data so that bias resulted from opportunistic sampling in the clinic can be reduced during survival analysis.…”

“… 14 Irurzun‐Arana et al identified dynamic change in lactate dehydrogenase (LDH) as the most significant predictor of OS in melanoma. 17 Such analysis has so far not been reported for RPS.…”

“… 6 , 10 , 11 , 12 , 13 However, recent studies have demonstrated the value of incorporating continuous progression profiles of various biological or physical biomarkers, apart from time‐invariant covariates, in oncologic survival analysis, including but not limited to tumor size, therapeutic targets, and disease‐specific clinical variables. 14 , 15 , 16 , 17 , 18 , 19 This calls for development of paradigms linking quantitative models describing time‐varying biomarkers to models predicting clinical outcomes. 20 , 21 Longitudinal tumor size following treatment with durvalumab was linked to OS of urothelial carcinoma using a population tumor kinetic model plus a time‐to‐event (TTE) model.…”

“…The existing prognostic analysis of RPS focused on clinicopathologic and treatment characteristics, such as age, tumor grade, histologic subtype, extent of resection, etc., mainly using semiparametric methodologies 6,10–13 . However, recent studies have demonstrated the value of incorporating continuous progression profiles of various biological or physical biomarkers, apart from time‐invariant covariates, in oncologic survival analysis, including but not limited to tumor size, therapeutic targets, and disease‐specific clinical variables 14–19 . This calls for development of paradigms linking quantitative models describing time‐varying biomarkers to models predicting clinical outcomes 20,21 .…”

Impact of Circulating and Tissue Biomarkers in Adjuvant and Neoadjuvant Therapy for High-Risk Melanoma: Ready for Prime Time?

“…Few biomarkers have been specifically evaluated for the prediction of disease relapse in patients undergoing surgery for stage II and III melanoma. A retrospective study investigated the role of circulating biomarkers (serum LDH, melanoma-inhibiting activity [MIA], and calcium binding protein S100B) and their dynamic changes over time, with OS in melanoma patients treated with adjuvant interferon (IFN)-α2b [ 40 ]. In this analysis, all of the above biomarkers were significantly related to survival outcomes, and the dynamic change of LDH was the most significant predictor of OS.…”