Mutational loss of CDKN2A (encoding p16INK4A) tumor-suppressor function is a key genetic step that complements activation of KRAS in promoting the development and malignant growth of pancreatic ductal adenocarcinoma (PDAC). However, pharmacologic restoration of p16INK4A function with inhibitors of CDK4 and CDK6 (CDK4/6) has shown limited clinical efficacy in PDAC. Here, we found that concurrent treatment with both a CDK4/6 inhibitor (CDK4/6i) and an ERK–MAPK inhibitor (ERKi) synergistically suppresses the growth of PDAC cell lines and organoids by cooperatively blocking CDK4/6i-induced compensatory upregulation of ERK, PI3K, antiapoptotic signaling, and MYC expression. On the basis of these findings, a Phase I clinical trial was initiated to evaluate the ERKi ulixertinib in combination with the CDK4/6i palbociclib in patients with advanced PDAC (NCT03454035). As inhibition of other proteins might also counter CDK4/6i-mediated signaling changes to increase cellular CDK4/6i sensitivity, a CRISPR-Cas9 loss-of-function screen was conducted that revealed a spectrum of functionally diverse genes whose loss enhanced CDK4/6i growth inhibitory activity. These genes were enriched around diverse signaling nodes, including cell-cycle regulatory proteins centered on CDK2 activation, PI3K–AKT–mTOR signaling, SRC family kinases, HDAC proteins, autophagy-activating pathways, chromosome regulation and maintenance, and DNA damage and repair pathways. Novel therapeutic combinations were validated using siRNA and small-molecule inhibitor–based approaches. In addition, genes whose loss imparts a survival advantage were identified (e.g., RB1, PTEN, FBXW7), suggesting possible resistance mechanisms to CDK4/6 inhibition. In summary, this study has identified novel combinations with CDK4/6i that may have clinical benefit to patients with PDAC. Significance: CRISPR-Cas9 screening and protein activity mapping reveal combinations that increase potency of CDK4/6 inhibitors and overcome drug-induced compensations in pancreatic cancer.

Artificial intelligence and emerging data science techniques are being leveraged to interpret medical image scans. Traditional image analysis relies on visual interpretation by a trained radiologist, which is time-consuming and can, to some degree, be subjective. The development of reliable, automated diagnostic tools is a key goal of radiomics, a fast-growing research field which combines medical imaging with personalized medicine. Radiomic studies have demonstrated potential for accurate lung cancer diagnoses and prognostications. The practice of delineating the tumor region of interest, known as segmentation, is a key bottleneck in the development of generalized classification models. In this study, the incremental multiple resolution residual network (iMRRN), a publicly available and trained deep learning segmentation model, was applied to automatically segment CT images collected from 355 lung cancer patients included in the dataset “Lung-PET-CT-Dx”, obtained from The Cancer Imaging Archive (TCIA), an open-access source for radiological images. We report a failure rate of 4.35% when using the iMRRN to segment tumor lesions within plain CT images in the lung cancer CT dataset. Seven classification algorithms were trained on the extracted radiomic features and tested for their ability to classify different lung cancer subtypes. Over-sampling was used to handle unbalanced data. Chi-square tests revealed the higher order texture features to be the most predictive when classifying lung cancers by subtype. The support vector machine showed the highest accuracy, 92.7% (0.97 AUC), when classifying three histological subtypes of lung cancer: adenocarcinoma, small cell carcinoma, and squamous cell carcinoma. The results demonstrate the potential of AI-based computer-aided diagnostic tools to automatically diagnose subtypes of lung cancer by coupling deep learning image segmentation with supervised classification. Our study demonstrated the integrated application of existing AI techniques in the non-invasive and effective diagnosis of lung cancer subtypes, and also shed light on several practical issues concerning the application of AI in biomedicine.

Tissues are complex structures composed of different cell types, each of which present specific functions and characteristics. To better understand and measure the effect of tumor cell enrichment on protein pathway profiling and drug target activation measurements, the signaling activation portraits of laser capture microdissected (LCM) cancer epithelium and tumor stroma were compared with patient-matched whole-tissue specimens from 53 primary colorectal cancer samples. Microdissected material and whole-tissue lysate from contiguous cryostat sections were subjected to reverse-phase protein microarray analysis to determine the level of phopshorylation and expression of 75 different proteins known to be involved in cancer progression. The results revealed distinct differences in the protein activation portraits of cancer epithelium and stroma. Moreover, we found that the signaling activation profiles of the undissected whole-tissue specimens are profoundly different from the matched LCM material. Attempts to rescale the undissected pathway information based on percent endogenous tumor epithelium content were unsuccessful in recapitulating the LCM tumor epithelial signatures. Analysis of epidermal growth factor receptor phosphorylation and COX2 expression in these same sample sets revealed wholesale differences in the rank ordering of patient determination when LCM was compared with undissected samples. On the basis of these data, we conclude that accurate protein pathway activation status, which is under evaluation as a basis for patient selection and stratification for personalized therapy, must include upfront cellular-enrichment techniques such as LCM to generate accurate drug target activation status.

Endocrine therapy (ET) in combination with CDK4/6 inhibition is routinely used as first-line treatment for HR+/HER2− metastatic breast cancer (MBC) patients. However, 30–40% of patients quickly develop disease progression. In this open-label multicenter clinical trial, we utilized a hypothesis-driven protein/phosphoprotein-based approach to identify predictive markers of response to ET plus CDK4/6 inhibition in pre-treatment tissue biopsies. Pathway-centered signaling profiles were generated from microdissected tumor epithelia and surrounding stroma/immune cells using the reverse phase protein microarray. Phosphorylation levels of the CDK4/6 downstream substrates Rb (S780) and FoxM1 (T600) were higher in patients with progressive disease (PD) compared to responders (p = 0.02). Systemic PI3K/AKT/mTOR activation in tumor epithelia and stroma/immune cells was detected in patients with PD. This activation was not explained by underpinning genomic alterations alone. As the number of FDA-approved targeted compounds increases, functional protein-based signaling analyses may become a critical component of response prediction and treatment selection for MBC patients.