In This Issue

See article, p. 827.

Immune checkpoint blockade therapy targeting programmed cell death-1 (PD-1) or cytotoxic T lymphocyte–associated protein 4 (CTLA4) has been efficacious against a number of cancer types, particularly metastatic melanoma. However, the majority of patients who receive monotherapy regimens of immune checkpoint blockade do not exhibit durable long-term responses. To identify predictive biomarkers of response and putative mechanisms of resistance to immune checkpoint blockade, Chen, Roh, and colleagues performed immune profiling, via a 12-marker immunohistochemistry panel, and targeted gene expression profiling (GEP) on a set of longitudinal tumor samples from patients with metastatic melanoma who initially received anti-CTLA4 treatment and subsequently received anti–PD-1 therapy. Although clear predictive markers in pretreatment tumor biopsies were not identified, favorable immune signatures in early on-treatment tumor biopsies were highly predictive of response, particularly to anti–PD-1 therapy. GEP in longitudinal tumor samples identified potential mechanisms of therapeutic response and resistance and revealed mechanistic differences between CTLA4 blockade and PD-1 blockade. Nonresponders to anti–PD-1 therapy exhibited defects in interferon signaling, as well as defects in antigen processing and presentation. Together, these results provide evidence that early on-treatment assessment of tumor biopsies may help predict response to immune checkpoint blockade, and also provide insight into mechanisms of therapeutic resistance that are potentially actionable.

See article, p. 838.

Aberrant activation of FGFRs occurs in multiple tumor types, which has led to the development of FGFR inhibitors. However, it is unclear whether distinct activating genetic events similarly confer sensitivity to FGFR inhibition. Pearson and colleagues screened 341 patients with advanced solid tumors by FISH for FGFR gene amplification, and patients with FGFR1-amplified breast cancer and patients with FGFR2-amplified gastroesophageal cancer were subsequently treated with the pan-FGFR inhibitor AZD4547. Confirmed responses were observed in 1 of 8 FGFR1-amplified patients (12.5%) and 3 of 9 FGFR2-amplified patients (33%). All FGFR2-amplified tumors that responded to AZD4547 had high copy-number amplification and corresponding high-level expression of FGFR2. Elevated FGFR2 copy number was detected in the plasma DNA of all 3 responding patients, but not in patients with low-level amplification of FGFR2, suggesting that plasma DNA screening can detect high copy-number tumors. Consistent with the clinical results, high copy-number FGFR2-amplified cell lines were more sensitive to AZD4547 than those with lower copy-number amplification. FGFR1- and FGFR2-amplified cells were dependent on FGFR-mediated MAPK signaling, whereas high-level FGFR2-amplified cells also exhibited a unique dependence on PI3K. FGFR2 enhanced AKT phosphorylation in highly amplified cells via PI3K activation by multiple receptor tyrosine kinases, altogether revealing a distinct oncogene addiction mechanism relying on FGFR-mediated PI3K–AKT signaling. The finding that FGFR inhibitors may be more effective in tumors with high-level FGFR amplification has the potential to guide future clinical development of FGFR inhibitors and other drugs targeting amplified receptor tyrosine kinases.

See article, p. 852.

The risk of developing pancreatic ductal adenocarcinoma (PDAC) is elevated in obese individuals, and obesity is associated with a poor prognosis. PDAC is highly desmoplastic and characterized by activation of stromal pancreatic stellate cells (PSC) and accumulation of extracellular matrix. Obesity is also a predesmoplastic condition, and promotes fat accumulation and inflammation in the normal pancreas. Incio and colleagues hypothesized that obesity would promote adipocyte accumulation in PDAC and create a proinflammatory, profibrotic microenvironment to stimulate tumor progression. Consistent with this hypothesis, in multiple mouse models of PDAC, obesity resulted in enhanced tumor progression, and tumors in obese mice exhibited enlarged adipocytes, activated PSCs, and increased collagen I deposition, indicating a fibrotic adipose microenvironment. Obesity increased the inflammatory response, including production of the cytokine IL1β, which recruited tumor-associated neutrophils to activate PSCs, promoting tumor progression. Desmoplasia reduces blood vessel perfusion, and, consequently, in PDAC mice, obesity led to reduced tumor vascular perfusion and decreased 5-Fluorouracil (5-FU) uptake and efficacy. PSCs promote tumor desmoplasia via signaling through the angiotensin-II type-1 receptor (AT1), and inhibition of AT1 signaling with the receptor blocker losartan resulted in reduced obesity-induced desmoplasia, decreased tumor size, and enhanced chemosensitivity in obese mice. The preclinical findings were validated in patients with PDAC; adjuvant chemotherapy improved survival in normal-weight patients, but not obese patients. Together, these findings indicate that obesity-induced inflammation promotes tumor growth and chemoresistance. These effects may be reversed by antifibrotic drugs, such as AT1 inhibitors, which warrant clinical investigation in obese patients with PDAC.

See article, p. 870.

Infiltrating myeloid cells, including tumor-associated macrophages, promote pancreatic ductal adenocarcinoma (PDAC) growth and progression, in part via suppression of T-cell–mediated antitumor immune responses. However, the mechanisms underlying myeloid cell–driven immune suppression in PDAC remain incompletely understood. Kaneda and colleagues found that the PI3Kγ isoform was uniquely expressed by tumor-associated macrophages but not other infiltrating immune cells or pancreatic carcinoma cells. Genetic deletion of Pik3cg, which encodes murine PI3Kγ, augmented intratumoral accumulation of CD8⁺ T cells, impaired PDAC growth and metastasis, and prolonged survival in orthotopic and genetically engineered mouse models of PDAC. Consistent with this finding, pharmacologic blockade of PI3Kγ significantly suppressed PDAC progression and enhanced the response to standard-of-care chemotherapy with gemcitabine. Inhibition of PI3Kγ in macrophages diminished the expression of genes encoding immunosuppressive factors, such as Arg1, Il10, and Tgfb, but increased the expression of genes encoding immunostimulatory factors, including Ifng and Il12, and restored CD8⁺ T-cell recruitment, suggesting that PI3Kγ modulates transcriptional programming in tumor-associated macrophages to suppress antitumor T-cell responses. In addition, PI3Kγ promoted the expression of PDGF in macrophages, which was necessary for macrophage-driven stimulation of pancreatic cancer cell invasion and the desmoplastic response associated with PDAC. These findings identify PI3Kγ as an important regulator of immune suppression and tumor progression in PDAC and suggest that PI3Kγ inhibition may be an effective therapeutic strategy.

See article, p. 886.

Pancreatic ductal adenocarcinoma (PDAC) is marked by an abundant desmoplastic stroma composed of extracellular matrix proteins, cytokines, and various cell types including cancer-associated fibroblasts (CAF). However, whether this stroma is tumor-promoting or protective remains controversial, and the role of activated CAFs in PDAC progression is poorly understood. Waghray and colleagues assessed the heterogeneity of CAFs isolated from primary human PDACs and identified a subpopulation of functional cancer-associated mesenchymal stem cells (CA-MSC). CA-MSCs enhanced the proliferation and invasion of patient-derived human PDAC cells in vitro, and augmented pancreatic cancer growth, dissemination, and metastasis in an orthotopic model. The ability of CA-MSCs to promote PDAC growth and progression was dependent on secretion of the cytokine granulocyte macrophage colony-stimulating factor (GM-CSF), which signaled through its receptors on PDAC cells, resulting in elevated expression of markers of epithelial–mesenchymal transition (EMT) and increased cancer stem cell–like properties in pancreatic tumor cells. Depletion of GM-CSF expression in CA-MSCs suppressed CA-MSC–induced pancreatic cancer cell proliferation, invasion, and transendothelial migration, and blocked CA-MSC–driven tumor cell metastasis in vivo. These findings define a population of tumor-promoting MSCs within the PDAC microenvironment and establish GM-CSF as a mediator of mesenchymal–epithelial cross-talk and a potential therapeutic target in pancreatic cancer.

See article, p. 900.

CRISPR/Cas9-mediated genome editing and RNAi both allow for loss-of function screens to systematically identify essential cancer genes. Munoz and colleagues compared the utility of these screening methods to identify cancer vulnerabilities by targeting 2,722 genes in high-coverage proliferation-based screens using CRISPR/Cas9 or shRNA in 5 cancer cell lines. Approximately 2% to 3% of genes were found to be essential across cell lines by either approach. CRISPR/Cas9-based screens identified more essential genes than RNAi-based screens, suggesting that CRISPR had either a lower false-negative rate or a higher false-positive rate. Some genes that were not expressed were scored as essential in the CRISPR/Cas9 screen, indicating false-positive hits. These false-positive hits were associated with copy-number amplification at targeted loci, and were observed in aneuploid, but not diploid, cell lines. Even after the false-positive hits were removed, however, CRISPR identified more essential genes than RNAi, indicating that CRISPR screening allows for a more complete identification of genes required for tumor cell growth. Additionally, in CRISPR/Cas9 tiling arrays, in which all of the possible sgRNAs targeting 139 genes were tested, sgRNAs localized within conserved essential domains had the strongest effect, which might inform effective sgRNA library design. These findings indicate that CRISPR/Cas9 loss-of-function screens may identify more essential genes than RNAi screens, and suggest that highly conserved or amplified genomic regions influence the interpretation of CRISPR/Cas9 screen results and may guide sgRNA library design.

See article, p. 914.

The CRISPR/Cas9 system allows for site-specific genome editing via the introduction of double-strand DNA breaks by the Cas9 endonuclease. Aguirre, Meyers, and colleagues performed a genome-scale CRISPR/Cas9 loss-of-function screen to identify genotype- and phenotype-specific cancer cell vulnerabilities in 33 cancer cell lines. The screen revealed key oncogenic drivers as well as an unexpected correlation between increased genomic copy number and apparent essential genes. CRISPR/Cas9 targeting of genes within amplifications resulted in a reduction in cell proliferation and survival, even when the targeted genes were not expressed. This finding suggested that in amplified regions, a gene-independent effect of CRISPR/Cas9 targeting is present in addition to the direct effect of target gene deletion. Genes residing in high-level amplifications were identified among the most highly essential genes by the CRISPR/Cas9 screens, while analysis of RNAi screen results found fewer essential genes residing in copy-number amplifications. These results suggest that copy number must be known to avoid false-positive results in CRISPR/Cas9 screens. The gene-independent effects of CRISPR/Cas9 targeting on survival were correlated with the predicted number of Cas9-mediated cuts of target loci, and CRISPR/Cas9 targeting induced DNA damage and G2 cell-cycle arrest, which was proportional to the number of target loci. Altogether, these findings highlight that cancer cells are sensitive to site-specific DNA damage within amplified regions and indicate that copy number must be accounted for when interpreting CRISPR/Cas9 screens to avoid false-positive results.

Note: In This Issue is written by Cancer Discovery editorial staff. Readers are encouraged to consult the original articles for full details.