See article, p. 1358

  • The effects of APC mutations, which increase WNT signaling in colorectal cancer, can be reversed by TNKS inhibition.

  • New animal models, human cells, and organoids were used to circumvent issues with mouse colorectal cancer models.

  • Cases with different mutations in the same gene should be evaluated separately for therapeutic response.


Hyperactive WNT signaling is seen in most colorectal cancers, and inactivating mutations in the tumor suppressor adenomatous polyposis coli (APC)—a scaffold protein mediating the formation of the destruction complex (DC) that facilitates β-catenin degradation—is the cause in 80% of such cases. Restoring DC activity (and, thus, normal WNT signaling) in the context of inactivated APC is possible through pharmacologic inhibition of tankyrase (TNKS) 1 and 2, which are functionally redundant. Using APC-mutant animal models, human cells, and ex vivo organoid cultures, Schatoff and colleagues found that TNKS inhibition suppressed hyperactive WNT signaling and controlled tumor growth in vivo. However, whether TNKS inhibition was effective depended on the mechanism of APC disruption: APC mutants with truncations in the mutation cluster region were still able to regulate β-catenin and responded to TNKS blockade, whereas this was not the case when there were truncations earlier in the sequence. Truncations in the mutation cluster region are commonly observed in patients, whereas the earlier truncations are present in commonly used mouse models. Collectively, these results indicate that TNKS inhibition can restore control of WNT signaling in some APC-mutant cases and illustrate that different mutations in the same gene, even those causing the same phenotype (in this case, WNT hyperactivation), can respond differently to targeted therapies.

See article, p. 1372

  • Olaparib with temozolomide (OT) showed preliminary evidence of efficacy in small-cell lung cancer.

  • A co-clinical trial using patient-derived xenografts identified biomarkers for treatment response.

  • Further trials of OT or similar combinations using doses defined in this study are warranted.


The combination of poly (ADP-ribose) polymerase (PARP) inhibitors with DNA-damaging agents is a treatment prospect of interest in small-cell lung cancer (SCLC). Farago and colleagues conducted a single-arm phase I/II clinical trial of the FDA-approved PARP inhibitor olaparib with the DNA-alkylating drug temozolomide (OT) in 50 patients with relapsed SCLC. The primary objective for the phase I portion of the trial was to determine the dose for the phase II portion of the trial, which had the goal of assessing efficacy. The overall response rate (ORR) in the 48 evaluable patients was 41.7% (20/48), with 20 patients exhibiting a confirmed partial response and an additional 4 patients experiencing an unconfirmed partial response. The median duration of response was 4.3 months, and after a median follow-up period of 7.1 months, the median progression-free survival was 4.2 months, and the median overall survival was 8.5 months. Experiments using patient-derived xenograft models faithfully recapitulated clinical data from the patients. The PDX models revealed that sensitivity to standard first-line therapy with etoposide and platinum (EP) correlated with response to OT. Further, sensitivity to EP and OT was associated with increased transcription of inflammatory-response gene sets, whereas resistance to EP and OT was correlated with expression of MYC-regulated transcripts. The PDX models also defined possible biomarkers for response to OT, including CEACAM1 (encoding CD66B), OAS1, TNFSF10 (encoding TRAIL), and TGIF1. This study provides a basis for further trials of OT and possibly other combinations of PARP inhibitors with DNA-damaging agents.

See article, p. 1388

  • Biomarker-based treatment was associated with improved response in patients with gastric cancer.

  • Patients with MET amplifications receiving savolitinib exhibited the most promising responses.

  • The PD-L1 status of patients treated with selumetinib and docetaxel changed over time.


Recent studies have revealed heterogeneity in the genomic alterations found in gastric cancer. In the VIKTORY trial, Lee and colleagues classified 772 patients with metastatic gastric cancer who had undergone one round of cytotoxic chemotherapy into eight biomarker-based groups, when possible, and matched them with clinical trials accordingly. Compared with a cohort not assigned to trials based on biomarkers, patients matched with treatments based on biomarkers had improved response rates (median overall survival of 9.8 versus 6.9 months; median progression-free survival of 5.7 versus 3.8 months). After correcting for prognostic factors such as age, gender, number of organs involved, Epstein–Barr virus status, mismatch-repair status, and performance status, biomarker-based treatment still predicted better response. The most promising results were seen in the patient group with MET amplifications assigned to monotherapy with the reversible MET kinase inhibitor savolitinib, with an overall response rate of 50% (10/20 patients). In this group, higher MET copy number was associated with increased probability of response, a phenomenon that has also been observed in patients with gastric cancer with HER2 or EGFR amplifications treated with HER2- or EGFR-targeted therapies. Additionally, it was observed in a small group of patients that PD-L1 status changed with time following treatment with selumetinib and docetaxel, a finding that may have implications for use of PD-L1 as a predictive biomarker for immunotherapy in gastric cancer. Collectively, these results highlight the potential utility of biomarker-based treatment for gastric cancer, perhaps especially in cases associated with MET amplification.

See article, p. 1406

  • Langerhans cell histiocytosis (LCH) lesions exhibited cellular and molecular heterogeneity.

  • LCH cells were classified, including a proliferative compartment, defining cell types with specific properties.

  • The results support a developmental hierarchy in LCH lesions and provide insight into LCH progression.


Langerhans cell histiocytosis (LCH) is a rare, MAPK pathway–driven hematopoietic disorder that lies between traditionally defined cancers and inflammatory diseases and exhibits marked clinical heterogeneity. Through a single-cell transcriptomic analysis of biopsies from seven LCH lesions from different sites, Halbritter, Farlik, and colleagues developed a cellular and molecular characterization of LCH. In addition to LCH cells, multiple distinct clusters of immune cells were identified, including T cells, B cells, macrophages, and dendritic cells. LCH cells exhibited high expression of genes previously identified as being specifically upregulated in LCH (e.g., MMP9) as well as genes important for antigen presentation, genes for HLA-complex members, MYC-associated genes, genes associated with the cell cycle, and genes associated with DNA repair. These findings support the characterization of LCH as both an inflammatory disease and a cancer. The single-cell resolution of the dataset enabled analysis of the cellular and molecular heterogeneity among LCH cells, which revealed a developmental hierarchy within LCH lesions; specifically, fourteen distinguishable cell categories ranging from low to high differentiation status were represented in samples from all seven lesions. ATAC-seq experiments showed that there were characteristic patterns of chromatin accessibility for each LCH cell type, and specific gene-regulatory networks distinguished each cell type, supporting the observed developmental hierarchy of LCH cell types. Collectively, these findings provide evidence for a high degree of heterogeneity within LCH lesions and lay the groundwork for further studies of the developmental progression of LCH cells.

See article, p. 1422

  • Contact between healthy NK cells and myeloid leukemia cells triggered PD-L1 expression on NK cells.

  • High PD-L1 expression by NK cells in patients with AML was associated with complete remission.

  • PD-L1 expression in human NK cells was regulated by the PI3K/AKT signaling pathway.


Blockade of programmed death ligand 1 (PD-L1) with monoclonal antibodies has been successfully used to treat PD-L1+ tumors; however, for unknown reasons, PD-L1 tumors also sometimes respond to anti–PD-L1 therapy. Dong, Wu, Ma, Wang, and colleagues found that direct contact between myeloid leukemia cells and natural killer (NK) cells from healthy donors caused a marked increase in PD-L1 expression at both the RNA and protein level in the NK cells, and PD-L1 protein was present on the surface of the NK cells and secreted by the NK cells. High PD-L1 expression was characteristic of activated NK cells, and data from 79 patients with acute myeloid leukemia (AML) revealed that high expression of PD-L1 by NK cells was associated with an increased chance of complete remission (CR). Exposure to the anti–PD-L1 therapy atezolizumab enhanced the activity of NK cells, and in vivo experiments showed that mouse NK cells with high PD-L1 expression had increased antitumor effects. Surprisingly, this mechanism appeared to be PD-1 independent. In an orthotopic mouse model, atezolizumab increased the antitumor activity of PD-L1–expressing NK cells, prolonging survival of mice engrafted with lethal doses of human myeloid leukemia cells. Experiments on human NK cells revealed that PD-L1 expression was regulated by the PI3K/AKT signaling pathway. Together, these results provide new insight into the mechanisms behind PD-L1′s activity in cancer cells and suggest that anti–PD-L1 therapy in patients with AML with highly PD-L1–expressing NK cells at CR may be worth considering.

See article, p. 1438

  • Cancers frequently have mutations in histone bodies, not just in post-translationally modified tails.

  • The most common mutant histone (H2BE76K) incorporates into chromatin and destabilizes nucleosomes.

  • H2BE76K increases chromatin accessibility, alters gene expression, and increases proliferation.


Mutations that disrupt components of the gene-regulatory machinery, including mutations that affect post-translational modification of histone tail residues, are frequently found in cancers. By examining cancer genomics data from 41,738 patient samples representing all cancer types, Bennett and colleagues discovered a previously unrecognized class of cancer-associated mutations that affect the bodies of core histone proteins (H2A, H2B, H3, and H4). The most common mutation created a glutamate-to-lysine residue substitution at position 76 of H2B. In vitro experiments suggested that nucleosomes with H2BE76K are unstable. Replacement of one of Saccharomyces cerevisiae's H2B genes with one encoding H2BE79K (equivalent to human H2BE76K) reduced nucleosome stability and increased chromatin accessibility. A reduction in nucleosome-mediated gene repression was also observed. Histone octamer formation was disrupted by H2BE76K, resulting in defects in nucleosome structure. In normal mammary epithelial cells, expression of H2BE76K and the oncogene PIK3CAH1047R caused increased proliferation relative to cells with PIK3CAH1047R alone, and H2BE76K expression caused altered expression of genes related to differentiation, apoptosis, proliferation, migration, and signaling. H2BE76K was also demonstrated to incorporate into chromatin in these cells, and its expression increased chromatin accessibility, particularly in normally transcriptionally inactive chromatin regions. An increase in accessibility at the promoters of more than 3,200 genes and increased expression of those genes was also observed. This work provides the basis for future studies on the biological effects of histone body mutations and the degree to which they may drive oncogenesis.

See article, p. 1452

  • Recurrent and lineage-specific mutations were found in XPO1 in patients with cancer.

  • XPO1E571K promoted oncogenesis in vitro and in vivo and altered nucleo-cytoplasmic trafficking.

  • XPO1 mutations may confer increased sensitivity to XPO1 inhibitors such as selinexor.


Exportin-1 (XPO1; also known as CRM1) is the primary nuclear-export receptor for proteins 40 kDa and larger. XPO1 is overexpressed in many cancers, and drugs that inhibit XPO1 are now FDA-approved for multiple myeloma and are in clinical trials for multiple other cancer types. Taylor and colleagues performed a genomic analysis of 42,793 patients with cancer—including 322 cancer types—and discovered highly recurrent and lineage-specific mutations in XPO1 at E571, D624, and R749; the latter two mutations had not previously been identified. Mouse xenotransplantation experiments revealed that XPO1E571K/WT cells exhibited increased growth. In a conditional knock-in mouse model, Xpo1E571K expression was associated with low body weight, splenomegaly, and B-cell proliferation and transformation, and Xpo1E571K promoted c-MYC– and BCL2-driven lymphomagenesis. Proteomic analysis of an XPO1E571K/WT human B-cell malignancy line identified many differentially exported proteins, including but not limited to members of the K63-ubiquitination, TLR4, and NFκB pathways. Structural modeling predicted that mutation of the negatively charged glutamic acid residue at position 571 to the positively charged lysine residue, as in XPO1E571K, would promote interaction with and export of proteins with negatively charged residues C terminal to their nuclear export sequences; sequence analysis of the proteomic data and in vitro nuclear-export assays confirmed this. Both in vitro and in vivo experiments revealed that the XPO1E571K mutation was associated with increased sensitivity to treatment with the XPO1 inhibitor selinexor. These findings suggest that mutations that alter the functions of nucleo-cytoplasmic trafficking machinery promote tumorigenesis and support the investigation of targeted therapies for patients with XPO1 hotspot mutations.

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