In This Issue

See article, p. 526.

Approximately 7% of estrogen receptor–positive (ER⁺) breast cancers harbor mutations in AKT1 (most often E17K) that cause PI3K-pathway activation. However, due to the rarity of this genomically defined breast cancer subgroup, the potential prognostic and therapeutic relevance of AKT1^E17K is difficult to ascertain. Using data from the American Association for Cancer Research's genomics registry and data-sharing consortium Project Genomics Evidence Neoplasia Information Exchange (GENIE), Smyth, Zhou, Nguyen, and colleagues performed a retrospective evaluation of 455 patients with AKT1^E17K-mutant or wild-type ER⁺ metastatic breast cancer. Patients had similar baseline characteristics, received similar therapies, and had comparable overall survival, but the duration of treatment (DOT)—used as a surrogate for treatment benefit—with the mTOR inhibitor everolimus was longer in the AKT1^E17K-mutant cohort than in the wild-type cohort. Notably, though, the DOT with mTOR-inhibitor treatment for patients with PI3K-pathway activation through means other than AKT1 mutation was the same as the DOT for patients without PI3K-pathway activation. One caveat of this work is that controlling for some variables is not possible when multiple studies are pooled; for instance, this study lacked data on patient performance and comorbid conditions. This limitation is balanced by a strength of this type of study; namely, in this case, combining data from several studies allowed the impact of a rare mutation—which individual studies were statistically underpowered to detect—to be evaluated. This work highlights both the promise and challenges of the use of real-world datasets and provides a framework for future investigations of rare mutations.

See article, p. 536.

For patients with acute myeloid leukemia (AML), treatment with the specific BCL2 inhibitor venetoclax plus the hypomethylating agent azacitidine causes remission in 70% of cases, but resistance or relapse can occur. In a retrospective analysis of 100 previously untreated patients with AML who received venetoclax plus azacitidine, Pei and colleagues found that myeloid differentiation status was a strong predictor of response, with patients who had monocytic disease being much less likely to respond to the combination compared with patients with less differentiated (primitive) forms of AML. In vitro experiments demonstrated that, compared with primitive AML, monocytic AML is characterized by leukemic stem cells that are much more resistant to the combination treatment. Monocytic AML also exhibited a transcriptional profile that was distinct from that of primitive AML and indicative of increased oxidative phosphorylation activity. Additionally, morphologic maturation of monocytic AML cells correlated with a loss of BCL2 expression, mimicking a process also seen during normal monocytic development. To compensate for the decrease in oxidative phosphorylation that would be expected with BCL2 loss, monocytic AML cells appeared to rely on the related protein MCL1. Analysis of relapsed clinical samples revealed that treatment with venetoclax plus azacitidine selected for monocytic disease arising from preexisting monocytic subclones that had dedifferentiated to adopt a more stemlike phenotype, although they retained myeloid characteristics and dependence on MCL1. Collectively, these findings illustrate a mechanism of acquired resistance to venetoclax plus azacitidine in AML and suggest that targeting MCL1 may be of interest in relapsed disease.

See article, p. 552.

CAR-T cells targeting the B-cellsurface protein CD19 have revolutionized treatment of some B-cell malignancies. However, the mechanisms responsible for lack of response, termed primary resistance, are poorly understood, a problem Singh, Lee, and colleagues addressed using a genome-wide CRISPR/Cas9-based screen in a CD19⁺ human acute lymphoblastic leukemia (ALL) cell line. This screen demonstrated that death receptor signaling in ALL cells may underlie primary resistance to CAR T cells, a finding verified in multiple other models of B-cell malignancy. Experiments using CD19⁺ ALL cells with loss of one of the two most enriched genes in the CRISPR/Cas9 screen, BID (encoding a death agonist), revealed that impaired death receptor signaling in ALL cells led to CAR T-cell dysfunction that worsened with longer exposures to ALL. Additionally, transcriptomic and epigenomic analyses showed that the dysfunctional CAR T cells exhibited a phenotype reminiscent of exhausted conventional T cells. Evaluation of gene expression in samples from clinical trials of the CD19-targeting CAR T-cell therapy tisagenlecleucel indicated that expression of death receptor genes was associated with durable response, CAR T-cell expansion and persistence, and patient survival. Further, single-cell RNA-sequencing analyses of clinical CAR T samples revealed that T cells from a nonresponding patient had substantially higher expression of exhaustion-associated genes as compared with T cells from a patient who had a durable complete response. These results demonstrate that CAR T cells exposed to ALL with dysregulated death receptor signaling develop a dysfunctional phenotype, and that both dysregulated death receptor signaling and a dysfunctional T-cell phenotype are associated with poor outcomes after CAR T-cell therapy.

See article, p. 568.

In B-progenitor acute lymphoblastic leukemia, disease recurrence is a major cause of mortality. There is some evidence that minor subclones present at diagnosis may give rise to relapse, but the reasons these subclones survive treatment and when during the course of treatment they become drug tolerant are unknown. Using a limiting-dilution assay in which purified leukemic blasts from samples taken at diagnosis and relapse were transplanted intrafemorally into immunosuppressed mice, Dobson and colleagues found evidence that leukemia-initiating cells that initiated the grafts were genetically diverse, often including all the clones present at diagnosis. Notably, these experiments with patient-derived xenografts revealed the engraftment of subclones latent at diagnosis. Further, analysis of the repopulation kinetics of subclones from diagnostic samples confirmed that relapse-initiating subclones were already present at diagnosis. Additionally, functional and genotyping experiments showed that relapse-initiating subclones present at diagnosis presented with distinct characteristics—including immunophenotype, repopulation potential, proliferation capability, and migration properties—even prior to chemotherapy exposure. Subclones present at diagnosis also showed different responses to chemotherapy, with relapse-initiating subclones being more resistant at baseline to commonly used chemotherapeutic drugs such as l-asparaginase or vincristine. Finally, transcriptional analyses revealed that relapse-initiating subclones present at diagnosis were enriched for transcription of genes relevant to chromatin remodeling, mitochondrial metabolism, proteostasis, and stemness. Collectively, this work establishes that relapse-fated subclones are already present at diagnosis, prior to undergoing the selective pressures imposed by chemotherapy, and reveals their distinct characteristics, which may be further investigated to uncover therapeutic vulnerabilities.

See article, p. 588.

Oncogenic KRAS mutations drivepancreatic ductal adenocarcinoma (PDAC) development, and Kras-mutant mice are often used to model PDAC. However, oncogenic Kras mutations alone are generally not sufficient to cause PDAC development—additional contributors, such as Myc mutations, are required. Using a Kras^G12D-mutant mouse model of PDAC in which Myc expression in the pancreas could be toggled on and off, Sodir and colleagues assessed the roles of Myc activation in the progression from pancreatic intraepithelial neoplasia (PanIN) to PDAC. In this model, Myc activation resulted in immediate initiation of PDAC development from early-stage PanIN lesions. Further, Myc activation induced the transcription of several signaling molecule–encoding genes, including Gas6, which encodes a ligand of the tyrosine protein kinase receptor AXL. Notably, inhibition of GAS6–AXL signaling substantially reduced the MYC-driven progression of PanIN lesions, lessened tumor burden, and abrogated the development of stromal aberrations characteristic of PDAC. Interestingly, following the establishment of MYC-induced PDAC, Myc inactivation led to rapid regression of even highly advanced tumors, indicating that continuous MYC activation is required not only for PDAC initiation, but also for PDAC maintenance in this model. Mechanistically, Myc deactivation was associated with a transition of PDAC epithelial cells from a ductal type to their original acinar type, and tumor cell death appeared to be ascribable to B cell–mediated recruitment of innate natural killer cells. In summary, this study demonstrates not only how Myc activation promotes mutant Kras–driven PDAC, but also how Myc deactivation can reverse this transition.

See article, p. 608.

Numerous studies have been dedicated to understanding thetumor-restraining effects of immunecells in the tumor microenvironment (TME); however, in some cases, tumor immune infiltrates appear to promote tumorigenesis. For example, some types of infiltrating CD4⁺ T cells, such as T_h2 cells, may contribute to inflammation and tumor growth by inducing and polarizing M1 macrophages, producing immunosuppressive M2 macrophages. Dey and colleagues found that expression of oncogenic KRAS or Kras mutants in patient pancreatic ductal adenocarcinomas (PDAC) and mouse PDAC cells, respectively, led to upregulation of the genes encoding type I cytokine receptors. Experiments using an orthotopic mouse model of mutant Kras–driven PDAC indicated that IL4Rα, but not IL2Rγ, promoted PDAC tumorigenesis in vivo. In PDAC cell lines and organoids, cell proliferation was driven by the cytokines IL4 and IL13, likely by promoting JAK–STAT-mediated upregulation of Myc. Specifically, increased MYC levels triggered metabolic programming to favor glycolysis- and tricarboxylic acid cycle–dependent energy production. Notably, IL4 and IL13 appeared to be supplied by T_h2 cells in the TME, implying a tumorigenic role for these infiltrating T lymphocytes in PDAC, and further experiments indicated that IL4 and IL13 may act in the early stages of tumor initiation and growth. This work lays out a mechanism by which TME-supplied IL4 and IL13 promote tumorigenesis in Kras-mutant PDAC and suggests that targeting the underlying pathway may be worth pursuing.

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