Automated tabulation of clinical trial results: A joint entity and relation extraction approach with transformer-based language representations.

Evidence-based medicine, the practice in which healthcare professionals refer to the best available evidence when making decisions, forms the foundation of modern healthcare. However, it relies on labour-intensive systematic reviews, where domain specialists must aggregate and extract information from thousands of publications, primarily of randomised controlled trial (RCT) results, into evidence tables. This paper investigates automating evidence table generation by decomposing the problem across two language processing tasks: named entity recognition, which identifies key entities within text, such as drug names, and relation extraction, which maps their relationships for separating them into ordered tuples. We focus on the automatic tabulation of sentences from published RCT abstracts that report the results of the study outcomes. Two deep neural net models were developed as part of a joint extraction pipeline, using the principles of transfer learning and transformer-based language representations. To train and test these models, a new gold-standard corpus was developed, comprising over 550 result sentences from six disease areas. This approach demonstrated significant advantages, with our system performing well across multiple natural language processing tasks and disease areas, as well as in generalising to disease domains unseen during training. Furthermore, we show these results were achievable through training our models on as few as 170 example sentences. The final system is a proof of concept that the generation of evidence tables can be semi-automated, representing a step towards fully automating systematic reviews.

Evaluating complete remission with partial hematologic recovery (CRh) as a response criterion in myelodysplastic syndromes (MDS).

Myelodysplastic syndromes (MDS) treated with DNMTI therapy have responses according to the 2006 IWG response criteria. CR responses have had the strongest association with OS. Recently, CR with partial hematologic recovery (CRh; i.e. blasts <5%, ANC > 500, platelets > 50) has been evaluated in AML, but its relevance is unknown in MDS. We identified adult patients with MDS treated with DNMTIs. We assessed best overall response to therapy according to IWG 2006 criteria, and subsequently identified patients meeting CRh criteria from the subgroup with SD or mCR. We evaluated duration of therapy and overall survival according to response. We identified 311 patients with MDS who received treatment between 2007 and 2018. The median age at the time of therapy was 69 years (range 23-91). Median follow up was 60 months. According to IWG 2006, responses included CR (n = 43, 14%), PR (n = 2, 1%), mCR (n = 57, 18%), SD (n = 149, 48%) and PD (n = 60, 19%). 79 patients (25%) achieved HI. A total of 62 patients (20%) met CRh criteria leading to reclassification of mCR (now n = 26, 8%) or SD (now n = 118, 38%). Patients achieving CR had similar time on therapy (median 8.1mo) compared to CRh (median 6mo, HR 1.4, 95% CI 0.9-2.0), and longer than other responses (p < 0.001). OS varied according to response; median OS was similar between CR (23.3mo) and CRh (25mo, HR 1.28 [0.79-2.08]), which was longer than those with mCR (17.2mo, HR 1.71 [0.96-3.05]), SD (16.3mo, HR 1.61 [1.04-2.48]), and PD (8.7mo, HR 3.04 [1.91-4.83]) (p < 0.001). OS associations with CR/CRh were confirmed in multivariable analysis accounting for allogeneic transplant. MDS patients who achieve a CRh response had similar survival and duration on therapy as patients who achieve CR response and superior to other IWG responses. These data support further evaluation of CRh into future response criteria and clinical trials.

Progenitor Hierarchy of Chronic Myelomonocytic Leukemia Identifies Inflammatory Monocytic-Biased Trajectory Linked to Worse Outcomes.

Myeloblast expansion is a hallmark of disease progression and comprises CD34+ hematopoietic stem and progenitor cells (HSPC). How this compartment evolves during disease progression in chronic myeloid neoplasms is unknown. Using single-cell RNA sequencing and high-parameter flow cytometry, we show that chronic myelomonocytic leukemia (CMML) CD34+ HSPC can be classified into three differentiation trajectories: monocytic, megakaryocyte-erythroid progenitor (MEP), and normal-like. Hallmarks of monocytic-biased trajectory were enrichment of CD120b+ inflammatory granulocyte-macrophage progenitor (GMP)-like cells, activated cytokine receptor signaling, phenotypic hematopoietic stem cell (HSC) depletion, and adverse outcomes. Cytokine receptor diversity was generally an adverse feature and elevated in CD120b+ GMPs. Hypomethylating agents decreased monocytic-biased cells in CMML patients. Given the enrichment of RAS pathway mutations in monocytic-biased cells, NRAS-competitive transplants and LPS-treated xenograft models recapitulated monocytic-biased CMML, suggesting that hematopoietic stress precipitates the monocytic-biased state. Deconvolution of HSPC compartments in other myeloid neoplasms and identifying therapeutic strategies to mitigate the monocytic-biased differentiation trajectory should be explored. SIGNIFICANCE: Our findings establish that multiple differentiation states underlie CMML disease progression. These states are negatively augmented by inflammation and positively affected by hypomethylating agents. Furthermore, we identify HSC depletion and expansion of GMP-like cells with increased cytokine receptor diversity as a feature of myeloblast expansion in inflammatory chronic myeloid neoplasms. This article is highlighted in the In This Issue feature, p. 476.

Asxl1 loss cooperates with oncogenic Nras in mice to reprogram the immune microenvironment and drive leukemic transformation.

Mutations in chromatin regulator ASXL1 are frequently identified in myeloid malignancies, in particular ∼40% of patients with chronic myelomonocytic leukemia (CMML). ASXL1 mutations are associated with poor prognosis in CMML and significantly co-occur with NRAS mutations. Here, we show that concurrent ASXL1 and NRAS mutations defined a population of CMML patients who had shorter leukemia-free survival than those with ASXL1 mutation only. Corroborating this human data, Asxl1-/- accelerated CMML progression and promoted CMML transformation to acute myeloid leukemia (AML) in NrasG12D/+ mice. NrasG12D/+;Asxl1-/- (NA) leukemia cells displayed hyperactivation of MEK/ERK signaling, increased global levels of H3K27ac, upregulation of Flt3. Moreover, we find that NA-AML cells overexpressed all the major inhibitory immune checkpoint ligands: programmed death-ligand 1 (PD-L1)/PD-L2, CD155, and CD80/CD86. Among them, overexpression of PD-L1 and CD86 correlated with upregulation of AP-1 transcription factors (TFs) in NA-AML cells. An AP-1 inhibitor or short hairpin RNAs against AP-1 TF Jun decreased PD-L1 and CD86 expression in NA-AML cells. Once NA-AML cells were transplanted into syngeneic recipients, NA-derived T cells were not detectable. Host-derived wild-type T cells overexpressed programmed cell death protein 1 (PD-1) and T-cell immunoreceptor with immunoglobulin and ITIM domains (TIGIT) receptors, leading to a predominant exhausted T-cell phenotype. Combined inhibition of MEK and BET resulted in downregulation of Flt3 and AP-1 expression, partial restoration of the immune microenvironment, enhancement of CD8 T-cell cytotoxicity, and prolonged survival in NA-AML mice. Our study suggests that combined targeted therapy and immunotherapy may be beneficial for treating secondary AML with concurrent ASXL1 and NRAS mutations.

Integrated Human and Murine Clinical Study Establishes Clinical Efficacy of Ruxolitinib in Chronic Myelomonocytic Leukemia.

PURPOSE: Chronic myelomonocytic leukemia (CMML) is a rare leukemia characterized by peripheral monocytosis with no disease-modifying therapies. CMML cells are uniquely hypersensitive to granulocyte-macrophage colony-stimulating factor (GM-CSF) and robustly engraft in immunocompromised mice that secrete human cytokines. To leverage these unique biological features, we conducted an integrated human and murine study evaluating ruxolitinib, a JAK1/2 inhibitor that potently downregulates intracellular GM-CSF signaling. PATIENTS AND METHODS: A total of 50 patients with WHO-defined CMML were enrolled in this open-label, multi-institution phase I/II clinical study, with a ruxolitinib dose of 20 mg twice daily studied in phase II. In parallel, 49 patient-derived xenografts (PDX) derived from 13 study participants were generated and randomized to receive ruxolitinib or vehicle control. RESULTS: The most common grade 3/4 treatment-related toxicities observed were anemia (10%) and thrombocytopenia (6%). The clinical overall response rate was 38% by Myelodysplastic Syndrome/Myeloproliferative Neoplasm (MDS/MPN) International Working Group (IWG) criteria and 43% of patients with baseline splenomegaly achieved a spleen response. Profiling of cytokine levels and somatic mutations at baseline failed to identify predictive biomarkers. PDX models derived from screening samples of study participants recapitulated responses seen in humans, particularly spleen responses, and corroborated ruxolitinib's clinical efficacy in a randomized murine study not feasible in human trials. CONCLUSIONS: Ruxolitinib demonstrated clinical efficacy and an acceptable adverse event profile in patients with CMML, identifying a potential novel therapeutic in this rare malignancy. Furthermore, this study demonstrates proof of concept that PDX modeling can recapitulate responses of patients treated on clinical trial and represents a novel correlative study that corroborates clinical efficacy seen in humans.See related commentary by Shastri and Adrianzen-Herrera, p. 6069.

Genomic Landscape and Risk Stratification in Chronic Myelomonocytic Leukemia.

PURPOSE: The advent of next-generation sequencing has allowed for the annotation of a vast array of recurrent somatic mutations across human malignancies, ushering in a new era of precision oncology. Chronic myelomonocytic leukemia is recognized as a myelodysplastic/myeloproliferative neoplasm and displays heterogenous clinical and genetic features. Herein, we review what is currently understood regarding the genomic landscape of this disease and discuss how somatic mutations have impacted current risk stratification methods. RECENT FINDINGS: Genomic studies in chronic myelomonocytic leukemia have identified a characteristic spectrum of cytogenetic and molecular abnormalities. Chromosomal abnormalities are detected in ~30% of patients and somatic gene mutations in up to 90% of patients, most commonly in TET2, SRSF2, and ASXL1. While cytogenetic abnormalities have long been known to impact the prognosis of myeloid neoplasms, recent studies have identified that somatic mutations impact prognosis independent of cytogenetic and clinical variables. This is best exemplified by mutations in ASXL1, which have been uniformly associated with inferior survival. These findings have led to the development of three molecularly inspired prognostic models, in an attempt to more accurately prognosticate in the disease. Our understanding of the genomic landscape of chronic myelomonocytic leukemia continues to evolve, with somatic mutations demonstrating an expanding role in diagnosis, risk stratification, and therapeutic decision-making. Given these findings, molecular profiling by next-generation sequencing should be considered standard of care in all patients.

Baseline and serial molecular profiling predicts outcomes with hypomethylating agents in myelodysplastic syndromes.

Hypomethylating agents (HMAs) are widely used in the treatment of myelodysplastic syndromes (MDSs), yet identifying those patients unlikely to benefit remains challenging. We assessed response and overall survival (OS) in 247 patients molecularly profiled by next-generation sequencing (NGS) before first-line HMA therapy, and a subset of 108 patients were sequenced serially during treatment. The most common mutations included TP53 (33.1%), ASXL1 (19%), TET2 (16.5%), DNMT3A (14.1%), and SRSF2 (12.1%). The overall response rate was 42.1%, with the composite TET2-mutant/ASXL1 wild-type genotype representing the strongest predictor of response (overall response rate, 62.1%; complete remission rate, 34.5%). The median OS for the cohort was 15 months, and the number of mutations detected by NGS (hazard ratio [HR], 1.22; P = .02), as well as mutations in TP53 (HR, 2.33; P = .001) and EZH2 (HR, 2.41; P = .04) were identified as independent covariates associated with inferior OS in multivariable analysis. Serial molecular profiling revealed that clearance of TP53 mutations during HMA therapy was associated with superior OS (HR, 0.28; P = .001) and improved outcome in patients proceeding to allogeneic hematopoietic cell transplantation. These data support baseline molecular profiling by NGS in MDS patients treated with HMAs and provide novel observations of sequential profiling during therapy that provide particular value in TP53-mutated disease.

Outcomes of patients with chronic myelomonocytic leukaemia treated with non-curative therapies: a retrospective cohort study.

BACKGROUND: Approval of hypomethylating agents in patients with chronic myelomonocytic leukaemia is based on trials done in patients with myelodysplastic syndromes. We aimed to investigate whether hypomethylating agents provide a benefit in subgroups of patients with chronic myelomonocytic leukaemia compared with other treatments. METHODS: For this retrospective cohort study, data were retrieved between Nov 30, 2017, and Jan 5, 2019, from 38 centres in the USA and Europe. We included non-selected, consecutive patients diagnosed with chronic myelomonocytic leukaemia, who received chronic myelomonocytic leukaemia-directed therapy. Patients with acute myeloid leukaemia according to 2016 WHO criteria at initial diagnosis (ie, ≥20% blasts in the bone marrow or peripheral blood) or with unavailability of treatment data were excluded. Outcomes assessed included overall survival, time to next treatment, and time to transformation to acute myeloid leukaemia. Analyses were adjusted by age, sex, platelet count, and Chronic myelomonocytic leukaemia-Specific Prognostic Scoring System (CPSS). Patients were grouped by first received treatment with either hydroxyurea, hypomethylating agents, or intensive chemotherapy, and stratified by risk according to blast count, French-American-British subtype, CPSS, WHO 2016 subtype, and the eligibility criteria of the DACOTA trial (NCT02214407). FINDINGS: 949 patients diagnosed with chronic myelomonocytic leukaemia between April 13, 1981, and Oct 26, 2018, were included. Median follow-up was 23·4 months (IQR 11·5-42·3) from diagnosis and 16·2 months (6·6-31·6) from start of first-line treatment. 412 (43%) of 949 patients received hypomethylating agents as first treatment, 391 (41%) hydroxyurea, and 83 (9%) intensive chemotherapy. Adjusted median overall survival for patients treated with hydroxyurea versus hypomethylating agents was 15·6 months (95% CI 13·1-17·3) versus 20·7 months (17·9-23·4); hazard ratio (HR) 1·39 (1·17-1·65; p=0·0002) and 14·0 months (9·8-17·2) versus 20·7 months (17·9-23·4; HR 1·55 [1·16-2·05]; p=0·0027) for those treated with intensive chemotherapy versus hypomethylating agents. In patients with myeloproliferative chronic myelomonocytic leukaemia (myeloproliferative CMML), median overall survival was 12·6 months (10·7-15·0) versus 17·6 months (14·8-21·5; HR 1·38 [1·12-1·70]; p=0·0027) for patients treated with hydroxyurea versus hypomethylating agents, and 12·3 months (8·4-16·6) versus 17·6 months (14·8-21·5; HR 1·44 [1·02-2·03]; p=0·040) for intensive chemotherapy versus hypomethylating agents. Hypomethylating agents did not confer an overall survival advantage for patients classified as having lower-risk disease (ie, myelodysplastic chronic myelomonocytic leukaemia with <10% blasts, CMML-0, or lower-risk CPSS). INTERPRETATION: These data suggest hypomethylating agents as the preferred therapy for patients with higher-risk chronic myelomonocytic leukaemia and those with myeloproliferative CMML. Our findings also suggest that CPSS is a valuable tool to identify patients who are most likely to benefit from hypomethylating agents. Further evidence from prospective cohorts would be desirable. FUNDING: The Austrian Group for Medical Tumor Therapy.

Molecular genetics of MDS/MPN overlap syndromes.

The myelodysplastic/myeloproliferative neoplasms (MDS/MPN) are a heterogenous group of myeloid malignancies hallmarked by clinicopathologic features that overlap with myelodysplastic syndromes and myeloproliferative neoplasms. Formally recognized by the World Health Organization, this group includes the entities chronic myelomonocytic leukemia, juvenile myelomonocytic leukemia, atypical chronic myeloid leukemia, MDS/MPN with ring sideroblasts and thrombocytosis and MDS/MPN, unclassifiable. Advancements in next generation sequencing have begun to unravel the molecular underpinnings of these diseases, identifying an array of recurrently mutated genes involved in epigenetic regulation, RNA splicing, transcription, and cell signaling. Despite molecular overlap with other myeloid malignancies, each entity displays a unique spectrum of somatic mutations supporting their unique pathobiology and clinical features. Importantly, molecular profiling is becoming an integral tool utilized in routine clinical practice. This review summarizes our current understanding of the molecular pathogenesis of overlap syndromes and details the impact of somatic mutations in diagnostic, prognostic, and therapeutic decision-making.

Targeting TP53 Mutations in Myelodysplastic Syndromes.

Mutations in TP53 are observed in ∼20% of patients with myelodysplastic syndromes (MDS), with increased frequency seen in patients with a complex karyotype and cases of therapy-related MDS. TP53 mutations represent perhaps the single greatest negative prognostic indicator in MDS. Inferior outcomes are demonstrated with all approved treatment approaches, although hypomethylating agents remain the standard frontline treatment option. Although outcomes with allogeneic hematopoietic stem cell transplant are poor, it remains the only potentially curative therapy. Novel agents are required to improve outcomes in this molecular subgroup, with therapies that directly target the mutant protein and immunotherapies demonstrating greatest potential.

Clinical, molecular, and prognostic correlates of number, type, and functional localization of TET2 mutations in chronic myelomonocytic leukemia (CMML)-a study of 1084 patients.

Loss-of-function TET2 mutations (TET2MT) are frequent early clonal events in myeloid neoplasms and are thought to confer a fitness advantage to hematopoietic precursors. This large, multi-institutional study (n = 1084), investigated the TET2 mutational landscape and prognostic implications of the number, type, and location of TET2MT and the epistatic relationship with other somatic events in chronic myelomonocytic leukemia (CMML). Nine hundred and forty-two TET2MT were identified in 604 (56%) patients, of which 710 (75%) were predicted to be truncating (involving the catalytic domain). Three hundred and sixteen (29%) patients had ≥1 TET2MT, with 28%, 1%, and 0.2% harboring 2, 3, and 5 mutations, respectively. In comparison to TET2WT, TET2MT patients were older in age, more likely to have dysplastic CMML, a higher number of co-occurring mutations, and lower-risk stratification. Importantly, TET2MT were associated with a survival advantage (49 vs. 30 months, p < 0.0001), especially in the context of multiple TET2MT (≥2; 57 months, p < 0.001), and truncating TET2MT (51 months, p < 0.001). In addition, the adverse prognostic impact of ASXL1MT was partially mitigated by concurrent TET2MT, with the ASXL1WT/TET2MT genotype having better outcomes and resulting in further risk stratification of ASXL1 inclusive CMML prognostic models, in comparison to ASXL1MT alone.

Current status and new treatment approaches in TP53 mutated AML.

Mutations in the essential tumor suppressor gene, TP53, are observed in only 5-10% of acute myeloid leukemia (AML) cases, but are highly associated with therapy-related AML and cases with complex karyotype. The mutational status of TP53 is a critical prognostic indicator, with dismal outcomes consistently observed across studies. Response rates to traditional cytotoxic chemotherapy are poor and long-term survival after allogeneic hematopoietic stem cell transplant is rare. Therapy with hypomethylating agents has resulted in a modest improvement in outcomes over intensive chemotherapy, but durable responses are seldom observed. In view of the intrinsic resistance to standard chemotherapies conferred by mutations in TP53, novel treatment approaches are required. In this review, we examine the current treatment landscape in TP53 mutated AML and discuss emerging therapeutic approaches currently under clinical investigation.

Current Management and Recent Advances in the Treatment of Chronic Myelomonocytic Leukemia.

OPINION STATEMENT: Chronic myelomonocytic leukemia (CMML) is an aggressive myeloid neoplasm in which treatment strategies with the capacity to improve survival are currently lacking. Clinical features are heterogeneous and although the overall prognosis is poor, survival can vary significantly between individuals. This reflects the need for an individualized treatment approach which incorporates accurate risk stratification. Though numerous prognostic scores exist, newer CMML-specific models incorporating molecular data should be favored. While asymptomatic, low-risk patients should be observed until their disease progresses, the majority of patients will require treatment. Due to a deficiency in treatments with disease-modifying capacity, any patient who requires treatment should be considered for enrollment in clinical trials evaluating novel therapeutic approaches. Allogeneic stem cell transplant (allo-SCT) remains the only current therapy with the potential to cure the disease and should be considered in most patients with intermediate- to high-risk disease. However, substantial risks are involved and, in part, because of advanced age at diagnosis, a minority of patients are candidates. Hypomethylating agents (HMAs) have become a preferred treatment approach, and should be used in those with cytopenias. Patients presenting with proliferative features can be treated with hydroxyurea to manage their symptoms and control leukocytosis, though HMAs can be incorporated as well, particularly in patients with higher risk disease. HMAs should also be considered in patients with a high burden of disease prior to proceeding with allo-SCT. Induction chemotherapy should be reserved for younger, healthy patients who have transformed to acute myeloid leukemia to induce remission prior to transplant. Supportive care utilizing transfusion support, erythropoiesis-stimulating agents, and infection prevention measures should be incorporated into the care of all patients.

An updated systematic review of lung chemo-radiotherapy using a new evidence aggregation method.

OBJECTIVE: The main treatment for locally-advanced, unresectable non-small cell lung cancer (NSCLC) is radical radiotherapy, with or without chemotherapy. Numerous trials have been performed, the results of which often disagree. Despite publication of several systematic reviews, there is not a consensus on practice, and outcomes are poor. To address this problem, we have applied a recently published evidence aggregation method to update the Cochrane systematic review (CSR) of lung chemo-radiotherapy. METHOD: We extracted data on 28 trials consisting of 4352 patients treated using 54 different regimens. We aggregated trials involving heterogeneous outcomes (both effects and side-effects). Our method identifies the evidence for the "best" treatment using preference criteria over outcomes and criteria concerning evidence quality. RESULT: Concurrent, platinum-based, conventionally fractionated chemo-radiotherapy was the most strongly supported option. 60 Gy in 30 fractions with concurrent cisplatin and vinblastine was superior under all combinations of preference criteria and evidence quality criteria that were considered. There was evidence for the use of concurrent and consolidation chemotherapy with conventionally-fractionated radiotherapy, but the choice of individual regimen was highly dependent on the choices for preference criteria over outcomes and for the criteria concerning evidence quality. DISCUSSION: Using evidence from the trials analysed in the CSR, together with some more recent studies, we identify a finer-grained recommendation than the CSR, though our findings are consistent with that of the CSR. Furthermore, we have identified individual regimens that may be optimal.

Aggregating evidence about the positive and negative effects of treatments.

OBJECTIVES: Evidence-based decision making is becoming increasingly important in healthcare. Much valuable evidence is in the form of the results from clinical trials that compare the relative merits of treatments. In this paper, we present a new framework for representing and synthesizing knowledge from clinical trials involving multiple outcome indicators. METHOD: The framework generates and evaluates arguments for claiming that one treatment is superior, or equivalent, to another based on the available evidence. Evidence comes from randomized clinical trials, systematic reviews, meta-analyses, network analyses, etc. Preference criteria over arguments are used that are based on the outcome indicators, and the magnitude of those outcome indicators, in the evidence. Meta-arguments attacks arguments that are based on weaker evidence. RESULTS: We evaluated the framework with respect to the aggregation of evidence undertaken in three published clinical guidelines that involve 56 items of evidence and 16 treatments. For each of the three guidelines, the treatment we identified as being superior using our method is a recommended treatment in the corresponding guideline. CONCLUSIONS: The framework offers a formal approach to aggregating clinical evidence, taking into account subjective criteria such as preferences over outcome indicators. In the evaluation, the aggregations obtained showed a good correspondence with published clinical guidelines. Furthermore, preliminary computational studies indicate that the approach is viable for the size of evidence tables normally encountered in practice.

Monoallelic silencing and haploinsufficiency in early murine intestinal neoplasms.

Studies of tumors from human familial adenomatous polyposis, sporadic colon cancer, and mouse and rat models of intestinal cancer indicate that the majority of early adenomas develop through loss of normal function of the Adenomatous polyposis coli (APC) gene. In murine models of familial adenomatous polyposis, specifically the multiple intestinal neoplasia mouse (Min) and the polyposis in the rat colon (Pirc) rat, most adenomas have lost their WT copy of the Apc gene through loss of heterozygosity by homologous somatic recombination. We report that large colonic adenomas in the Pirc rat have no detectable copy number losses or gains in genomic material and that most tumors lose heterozygosity only on the short arm of chromosome 18. Examination of early mouse and rat tumors indicates that a substantial subset of tumors shows maintenance of heterozygosity of Apc in genomic DNA, apparently violating Knudson's two-hit hypothesis. Sequencing of the Apc gene in a sampling of rat tumors failed to find secondary mutations in the majority of tumors that maintained heterozygosity of Apc in genomic DNA. Using quantitative allele-specific assays of Apc cDNA, we discovered two neoplastic pathways. One class of tumors maintains heterozygosity of Apc(Min/+) or Apc(Pirc/+) RNA expression and may involve haploinsufficiency for Apc function. Another class of tumors exhibits highly biased monoallelic expression of the mutant Apc allele, providing evidence for a stochastic or random process of monoallelic epigenetic silencing of the tumor suppressor gene Apc.