External Validation of a Dutch Predictive Nomogram for Complete Response to T-VEC in an Independent American Patient Cohort

Purpose: Talimogene Laherparepvec (T-VEC) is a modified herpes simplex virus type-1 used as intralesional immunotherapy in stage IIIB-IVM1a melanoma patients. Recently, Stahlie et al. published a predictive model for complete response (CR) to T-VEC. This study was designed to validate this model externally in an independent, American patient cohort.

Methods: In total, 71 stage IIIB-IVM1a melanoma patients treated with T-VEC at Moffitt Cancer Center were included. A second nomogram was built incorporating the same predictive factors: tumor size (diameter of largest metastasis), type of metastases (cutaneous, subcutaneous and nodal), and number of metastases (cutoff: < 20 and > 20). Predictive accuracy was assessed through calculation of overall performance, discriminative ability, and calibration.

Results: The two cohorts were similar in many clinicopathologic factors and only differing in tumor mutational status and use of systemic therapy prior to T-VEC. In the validation cohort, 37 (52%) patients showed CR, 22 (31%) partial response (PR), 2 (5.6%) stable disease (SD), and 10 (15%) progressive disease (PD). Of those who demonstrated a CR, 16 (43%) recurred. Overall performance was good (0.164), and discriminative power resulted in fair discriminative ability (0.827). The calibration curve showed slight underestimation for predicted probabilities > 0.15 and slight overestimation <0.15.

Conclusions: The original model as well as the validation model show comparable and good predictive accuracy. The validation model reinforces the conclusion that for the best response to T-VEC, it should be used early on in the course of the disease, when the tumor burden is cutaneous with smaller diameter and fewer of metastases.