Pancreatic ductal adenocarcinoma (PDAC), one of the main pancreatic cancer types with one of the lowest survival rates, has historically been characterized into 2 subtypes, basal and classical. These subtypes often exist in a spectrum of distinct niches affected my tumor microenvironment with an unknown spatial sequence of events during cancer invasion. Söderqvist et al. analyzed the pathway of lobular invasion using tumor cell markers and microenvironmental regulation to distinguish distinct tumor cell phenotypes. The researchers used histological slides to detect lobular invasion and characterize them. Moreover, a panel of 6 markers were used to find a correlation with PDAC subtypes in the lobules compared to in the stroma, demonstrating classical type tumor cells are distinct in lobules and basal tumor phenotypes are primarily in the stroma. Machine learning was utilized to identify complex tissue structures in a fibrotic context; the markers were then used to label samples and determine acinar cells with morphological signs of ADM primarily make up cancer cells within the periphery of lobules, similar to chronic pancreatitis-like lobular injury. This is characterized by NGFR+ inflamed stroma in lobular injury, which contributes to the tumor microenvironment; however, there is little to no presence of NGFR within stroma tumor cells. The researchers demonstrated abundance of certain proteins in specific areas of invasive tumors, allowing for deeper analysis into tumor phenotypic markers. Söderqvist et al. concluded that microenvironments within invaded lobules often contain ADM and NGFR⁺/PDGFRα⁺ which creates a microanatomical niche for these invasive cells. Furthermore, genetically engineered mice demonstrated ADM and reprogramming acinar cells to be major cancer initiation and precursor events with high spatial proximity, with loss of certain expression transforming lobular stroma into desmoplastic. Specific mutations have been observed to show complexities within PDAC heterogeneity and specific tumor-epithelial interactions can modulate specific phenotypes.