Imbalance in radiographic progression proportion as a potential confounder in evaluating the survival benefit of an adding-on strategy in metastatic pancreatic cancer