Extremely high genetic diversity in a single tumor points to prevalence of non-Darwinian cell evolution. Proc Natl Acad Sci U S A. 2015;112:E6496-E6505. [PMID: 26561581 DOI: 10.1073/pnas.1519556112]

Natural Selection in Cancer Biology: From Molecular Snowflakes to Trait Hallmarks

Evolution by natural selection is the conceptual foundation for nearly every branch of biology and increasingly also for biomedicine and medical research. In cancer biology, evolution explains how populations of cells in tumors change over time. It is a fundamental question whether this evolutionary process is driven primarily by natural selection and adaptation or by other evolutionary processes such as founder effects and drift. In cancer biology, as in organismal evolutionary biology, there is controversy about this question and also about the use of adaptation through natural selection as a guiding framework for research. In this review, we discuss the differences and similarities between evolution among somatic cells versus evolution among organisms. We review what is known about the parameters and rate of evolution in neoplasms, as well as evidence for adaptation. We conclude that adaptation is a useful framework that accurately explains the defining characteristics of cancer. Further, convergent evolution through natural selection provides the only satisfying explanation both for how a group of diverse pathologies have enough in common to usefully share the descriptive label of "cancer" and for why this convergent condition becomes life-threatening.

The Global Relationship between Chromatin Physical Topology, Fractal Structure, and Gene Expression

Most of what we know about gene transcription comes from the view of cells as molecular machines: focusing on the role of molecular modifications to the proteins carrying out transcriptional reactions at a loci-by-loci basis. This view ignores a critical reality: biological reactions do not happen in an empty space, but in a highly complex, interrelated, and dense nanoenvironment that profoundly influences chemical interactions. We explored the relationship between the physical nanoenvironment of chromatin and gene transcription in vitro. We analytically show that changes in the fractal dimension, D, of chromatin correspond to simultaneous increases in chromatin accessibility and compaction heterogeneity. Using these predictions, we demonstrate experimentally that nanoscopic changes to chromatin D within thirty minutes correlate with concomitant enhancement and suppression of transcription. Further, we show that the increased heterogeneity of physical structure of chromatin due to increase in fractal dimension correlates with increased heterogeneity of gene networks. These findings indicate that the higher order folding of chromatin topology may act as a molecular-pathway independent code regulating global patterns of gene expression. Since physical organization of chromatin is frequently altered in oncogenesis, this work provides evidence pairing molecular function to physical structure for processes frequently altered during tumorigenesis.

Tumor evolution: Linear, branching, neutral or punctuated?

Intratumor heterogeneity has been widely reported in human cancers, but our knowledge of how this genetic diversity emerges over time remains limited. A central challenge in studying tumor evolution is the difficulty in collecting longitudinal samples from cancer patients. Consequently, most studies have inferred tumor evolution from single time-point samples, providing very indirect information. These data have led to several competing models of tumor evolution: linear, branching, neutral and punctuated. Each model makes different assumptions regarding the timing of mutations and selection of clones, and therefore has different implications for the diagnosis and therapeutic treatment of cancer patients. Furthermore, emerging evidence suggests that models may change during tumor progression or operate concurrently for different classes of mutations. Finally, we discuss data that supports the theory that most human tumors evolve from a single cell in the normal tissue. This article is part of a Special Issue entitled: Evolutionary principles - heterogeneity in cancer?, edited by Dr. Robert A. Gatenby.

Bioinformatics and Drug Discovery

Bioinformatic analysis can not only accelerate drug target identification and drug candidate screening and refinement, but also facilitate characterization of side effects and predict drug resistance. High-throughput data such as genomic, epigenetic, genome architecture, cistromic, transcriptomic, proteomic, and ribosome profiling data have all made significant contribution to mechanismbased drug discovery and drug repurposing. Accumulation of protein and RNA structures, as well as development of homology modeling and protein structure simulation, coupled with large structure databases of small molecules and metabolites, paved the way for more realistic protein-ligand docking experiments and more informative virtual screening. I present the conceptual framework that drives the collection of these high-throughput data, summarize the utility and potential of mining these data in drug discovery, outline a few inherent limitations in data and software mining these data, point out news ways to refine analysis of these diverse types of data, and highlight commonly used software and databases relevant to drug discovery.

Stability of Cross-Feeding Polymorphisms in Microbial Communities

Cross-feeding, a relationship wherein one organism consumes metabolites excreted by another, is a ubiquitous feature of natural and clinically-relevant microbial communities and could be a key factor promoting diversity in extreme and/or nutrient-poor environments. However, it remains unclear how readily cross-feeding interactions form, and therefore our ability to predict their emergence is limited. In this paper we developed a mathematical model parameterized using data from the biochemistry and ecology of an E. coli cross-feeding laboratory system. The model accurately captures short-term dynamics of the two competitors that have been observed empirically and we use it to systematically explore the stability of cross-feeding interactions for a range of environmental conditions. We find that our simple system can display complex dynamics including multi-stable behavior separated by a critical point. Therefore whether cross-feeding interactions form depends on the complex interplay between density and frequency of the competitors as well as on the concentration of resources in the environment. Moreover, we find that subtly different environmental conditions can lead to dramatically different results regarding the establishment of cross-feeding, which could explain the apparently unpredictable between-population differences in experimental outcomes. We argue that mathematical models are essential tools for disentangling the complexities of cross-feeding interactions.

An exactly solvable, spatial model of mutation accumulation in cancer

One of the hallmarks of cancer is the accumulation of driver mutations which increase the net reproductive rate of cancer cells and allow them to spread. This process has been studied in mathematical models of well mixed populations, and in computer simulations of three-dimensional spatial models. But the computational complexity of these more realistic, spatial models makes it difficult to simulate realistically large and clinically detectable solid tumours. Here we describe an exactly solvable mathematical model of a tumour featuring replication, mutation and local migration of cancer cells. The model predicts a quasi-exponential growth of large tumours, even if different fragments of the tumour grow sub-exponentially due to nutrient and space limitations. The model reproduces clinically observed tumour growth times using biologically plausible rates for cell birth, death, and migration rates. We also show that the expected number of accumulated driver mutations increases exponentially in time if the average fitness gain per driver is constant, and that it reaches a plateau if the gains decrease over time. We discuss the realism of the underlying assumptions and possible extensions of the model.

Measuring intratumor heterogeneity by network entropy using RNA-seq data

Intratumor heterogeneity (ITH) is observed at different stages of tumor progression, metastasis and reouccurence, which can be important for clinical applications. We used RNA-sequencing data from tumor samples, and measured the level of ITH in terms of biological network states. To model complex relationships among genes, we used a protein interaction network to consider gene-gene dependency. ITH was measured by using an entropy-based distance metric between two networks, nJSD, with Jensen-Shannon Divergence (JSD). With nJSD, we defined transcriptome-based ITH (tITH). The effectiveness of tITH was extensively tested for the issues related with ITH using real biological data sets. Human cancer cell line data and single-cell sequencing data were investigated to verify our approach. Then, we analyzed TCGA pan-cancer 6,320 patients. Our result was in agreement with widely used genome-based ITH inference methods, while showed better performance at survival analysis. Analysis of mouse clonal evolution data further confirmed that our transcriptome-based ITH was consistent with genetic heterogeneity at different clonal evolution stages. Additionally, we found that cell cycle related pathways have significant contribution to increasing heterogeneity on the network during clonal evolution. We believe that the proposed transcriptome-based ITH is useful to characterize heterogeneity of a tumor sample at RNA level.

Genetic profiling of hepatocellular carcinoma using next-generation sequencing

Hepatocellular carcinoma (HCC) is a highly heterogeneous disease, both clinically and from a molecular standpoint. The advent of next-generation sequencing technologies has provided new opportunities to extensively analyze molecular defects in HCC samples. This has uncovered major cancer driver genes and associated oncogenic pathways operating in HCC. More sophisticated analyses of sequencing data have linked specific nucleotide patterns to external toxic agents and defined so-called 'mutational signatures' in HCC. Molecular signatures, taking into account intra- and inter-tumor heterogeneity, and their functional validation could provide useful data to predict treatment response to molecular therapies. In this review we will focus on the current knowledge of deep sequencing in HCC and its foreseeable clinical impact.

Spatial and temporal clonal evolution during development of metastatic urothelial carcinoma

Patients with metastatic bladder cancer have a median survival of only 13-14 months. Precision medicine using targeted therapy may improve survival. Here we investigated spatial and temporal tumour evolution and tumour heterogeneity in order to evaluate the potential use of targeted treatment of metastatic bladder cancer. We performed a proof-of-concept study by whole exome sequencing of multiple tumour regions (n = 22) from three patients with metastatic bladder cancer. DNA from primary and metastatic tumour biopsies was analysed for mutations using Mutect and potential therapeutic targets were identified. We identified 256, 265 and 378 somatic mutations per patient, encompassing mutations with an estimated functional impact in 6-12 known disease driver genes per patient. Disease driver mutations present in all tumour regions could be identified in all cases, however, over time metastasis specific driver mutations emerged. For each patient we identified 6-10 potentially therapeutic targets, however very few targets were present in all regions. Low mutational allele frequencies were observed in most regions suggesting a complex mixture of different cancer cells with no spatial demarcation of subclones. In conclusion, primary bladder tumours and metastatic lesions showed heterogeneity at the molecular level, but within the primary tumour the heterogeneity appeared low. The observed lack of potential therapeutic targets common to all cancer cells in primary tumours and metastases emphasizes the challenges in designing rational targeted therapy solely based on analysis of the primary tumours.

Excess of mutational jackpot events in expanding populations revealed by spatial Luria-Delbrück experiments

The genetic diversity of growing cellular populations, such as biofilms, solid tumours or developing embryos, is thought to be dominated by rare, exceptionally large mutant clones. Yet, the emergence of these mutational jackpot events is only understood in well-mixed populations, where they stem from mutations that arise during the first few cell divisions. To study jackpot events in spatially structured populations, we track mutant clones in microbial populations using fluorescence microscopy and population sequencing. High-frequency mutations are found to be massively enriched in microbial colonies compared with well-shaken liquid cultures, as a result of late-occurring mutations surfing at the edge of range expansions. Thus, jackpot events can be generated not only when mutations arise early but also when they occur at favourable locations, which exacerbates their role in adaptation and disease. In particular, because spatial competition with the wild type keeps most mutant clones in a quiescent state, strong selection pressures that kill the wild type promote drug resistance.

Large mutant clones arising from early mutations in growing cell populations facilitate short-term evolution in microbes and in tumours. Here the authors analyse spatially expanding colonies, and show that large mutant clones can also arise late when they surf at expanding frontiers.

Multiregion sequencing reveals the intratumor heterogeneity of driver mutations in TP53-driven non-small cell lung cancer

Intratumor heterogeneity (ITH) in non-small cell lung cancer (NSCLC) may account for resistance after a period of targeted therapies because drugs destroy only a portion of tumor cells. The recognition of ITH helps identify high-risk patients to make effective treatment decisions. However, ITH studies are confounded by interpatient heterogeneity in NSCLC and a large amount of passenger mutations. To address these issues, we recruited NSCLC patients carrying TP53 mutations and selected driver mutations within recurrently mutated genes in NSCLC. A total of 12-paired normal-tumor tissues were subjected to whole-genome/whole-exome sequencing. From these, 367 non-silent mutations were selected as driver mutations and deeply sequenced in 61 intratumoral microdissections. We identified a universal prevalence of heterogeneity in all 12 tumors, indicating branched evolution. Although TP53 mutations were observed in single biopsy of all 12 tumors, most tumors consist of both TP53 mutated and non-mutated cells in separate regions within the same tumor. This suggests the late molecular timing of the acquisition of TP53 mutations; therefore, the detection of TP53 mutations in a single biopsy may simply not reflect the early malignant potential. In addition, we identified regions of loss of heterozygosity surrounding TP53 and CDKN2A mutations in tumor 711, which also exhibited heterogeneity in different regional samples. Because the ITH of driver mutations likely has clinical consequences, further efforts are needed to limit the impact of ITH and to improve therapeutic efficiency, which will benefit NSCLC patients receiving targeted treatments.

New paradigms in clonal evolution: punctuated equilibrium in cancer

Evolutionary theories are themselves subject to evolution. Clonal evolution - the model that describes the initiation and progression of cancer - is entering a period of profound change, brought about largely by technological developments in genome analysis. A flurry of recent publications, using modern mathematical and bioinformatics techniques, have revealed both punctuated and neutral evolution phenomena that are poorly explained by the conventional graduated perspectives. In this review, we propose that a hybrid model, inspired by the evolutionary model of punctuated equilibrium, could better explain these recent observations. We also discuss the conceptual changes and clinical implications of variable evolutionary tempos. Copyright © 2016 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

A saltationist theory of cancer evolution

A new study based on single-nucleus sequencing reports that triple-negative breast cancers acquire copy number aberrations in short punctuated bursts in the earliest stages of tumor evolution, rather than continuously and gradually, challenging prevailing models of tumor evolution.

Detection of single cell heterogeneity in cancer

Single cell heterogeneity has already been highlighted in cancer classification, diagnosis, and treatment. Recent advanced technologies have gained more ability to reveal the heterogeneity on single cell level. In this review, we listed various detection targets applied in single cell study, including tumor tissue cells, circulating tumor cells (CTCs), disseminated tumor cells (DTCs), circulating tumor DNA (ctDNA), cell-free DNA (cfDNA), and cancer stem cells (CSCs). We further discussed and compared detection methods using these detection targets in different fields to reveal single cell heterogeneity in cancer. We focused not only on the methods that have already been established and validated, but also on newly developed methods. In morphology and phenotype, the methods mainly included cell imaging and immune-staining. In genomics and proteomics, the main methods were single cell sequencing and single cell western blotting. Collectively, from using these methods, we can have a better understanding of the single cell variation, as well as what kind of variation it is and how the variation works. Our observations imply that study on single cell heterogeneity in cancer is an important step to precision medicine. The development of technologies in detection of single cell heterogeneity will be sure to improve the diagnosis and treatment in cancer.

Epithelial-to-mesenchymal plasticity of cancer stem cells: therapeutic targets in hepatocellular carcinoma

Hepatocellular carcinoma (HCC) remains one of the most common and lethal malignancies worldwide despite the development of various therapeutic strategies. A better understanding of the mechanisms responsible for HCC initiation and progression is essential for the development of more effective therapies. The cancer stem cell (CSC) model has provided new insights into the development and progression of HCC. CSCs are specialized tumor cells that are capable of self-renewal and have long-term repopulation potential. As they are important mediators of tumor proliferation, invasion, metastasis, therapy resistance, and cancer relapse, the selective targeting of this crucial population of cells has the potential to improve HCC patient outcomes and survival. In recent years, the role of epithelial-to-mesenchymal transition (EMT) in the advancement of HCC has gained increasing attention. This multi-step reprograming process resulting in a phenotype switch from an epithelial to a mesenchymal cellular state has been closely associated with the acquisition of stem cell-like attributes in tumors. Moreover, CSC mediates tumor metastasis by maintaining plasticity to transition between epithelial or mesenchymal states. Therefore, understanding the molecular mechanisms of the reprograming switches that determine the progression through EMT and generation of CSC is essential for developing clinically relevant drug targets. This review provides an overview of the proposed roles of CSC in HCC and discusses recent results supporting the emerging role of EMT in facilitating hepatic CSC plasticity. In particular, we discuss how these important new insights may facilitate rational development of combining CSC- and EMT-targeted therapies in the future.

The Greater Genomic Landscape: The Heterogeneous Evolution of Cancer

Results have historically shown a broad plasticity in the origin of tumors and their functions, with significant heterogeneity observed in both morphologies and functional capabilities. Largely unknown, however, are the mechanisms by which these variations occur and how these events influence tumor formation and behavior. Contemporary views on the origin of tumors focus mainly on the role of particular sets of driver transformations, mutational or epigenetic, with the occurrence of the observed heterogeneity as an accidental byproduct of oncogenesis. As such, we present a hypothesis that tumors form due to heterogeneous adaptive selection in response to environmental stress through intrinsic genomic sampling mechanisms. Specifically, we propose that eukaryotic cells intrinsically explore their available genomic information, the greater genomic landscape (GGL), in response to stress under normal conditions, long before the formation of a cancerous lesion. Finally, considering the influence of chromatin heterogeneity on the GGL, we propose a new class of compounds, chromatin-protective therapies (CPT), which target the physical variations in chromatin topology. In this approach, CPTs reduce the overall information space available to limit the formation of tumors or the development of drug-resistant phenotypes. Cancer Res; 76(19); 5605-9. ©2016 AACR.

Punctuated copy number evolution and clonal stasis in triple-negative breast cancer

Aneuploidy is a hallmark of breast cancer; however, our knowledge of how these complex genomic rearrangements evolve during tumorigenesis is limited. In this study we developed a highly multiplexed single-nucleus-sequencing method to investigate copy number evolution in triple-negative breast cancer patients. We sequenced 1000 single cells from 12 patients and identified 1–3 major clonal subpopulations in each tumor that shared a common evolutionary lineage. We also identified a minor subpopulation of non-clonal cells that were classified as: 1) metastable, 2) pseudo-diploid, or 3) chromazemic. Phylogenetic analysis and mathematical modeling suggest that these data are unlikely to be explained by the gradual accumulation of copy number events over time. In contrast, our data challenge the paradigm of gradual evolution, showing that the majority of copy number aberrations are acquired at the earliest stages of tumor evolution, in short punctuated bursts, followed by stable clonal expansions that form the tumor mass.

First-in-Human Phase 1 Trial of Agarose Beads Containing Murine RENCA Cells in Advanced Solid Tumors

PURPOSE

Agarose macrobeads containing mouse renal adenocarcinoma cells (RMBs) release factors, suppressing the growth of cancer cells and prolonging survival in spontaneous or induced tumor animals, mediated, in part, by increased levels of myocyte-enhancing factor (MEF2D) via EGFR-and AKT-signaling pathways. The primary objective of this study was to determine the safety of RMBs in advanced, treatment-resistant metastatic cancers, and then its efficacy (survival), which is the secondary objective.

METHODS

Thirty-one patients underwent up to four intraperitoneal implantations of RMBs (8 or 16 macrobeads/kg) via laparoscopy in this single-arm trial (FDA BB-IND 10091; NCT 00283075). Serial physical examinations, laboratory testing, and PET-CT imaging were performed before and three months after each implant.

RESULTS

RMBs were well tolerated at both dose levels (mean 660.9 per implant). AEs were (Grade 1/2) with no treatment-related SAEs.

CONCLUSION

The data support the safety of RMB therapy in advanced-malignancy patients, and the preliminary evidence for their potential efficacy is encouraging. A Phase 2 efficacy trial is ongoing.

Precision medicine for hepatocellular carcinoma: Perspectives and obstacles

Hepatocellular carcinoma (HCC) is one of the most common cancers in China. HCC patients have a poor prognosis due to the lack of effective drugs. The marketing of sorafenib in 2007 has terminated the history that there is no effective drug for HCC. Unfortunately, sorafenib prolongs the survival of advanced HCC patients by only 2-3 mo, and remains the unique systemic drug as no alternative effective agents have been demonstrated to be superior to sorafenib in treating HCC. Precision medicine, a novel concept and medicinal model, has recently emerged and been spreading globally, with the development of gene sequencing techniques, bioinformatics, big data and so on. Detecting, analyzing, verifying and utilizing the specific tumor biomarkers with the advanced technology have made it possible to apply "personalized and precision therapy" in the treatment of advanced HCC. In the present article we summarize the recent progress of HCC therapy under the guidance of precision medicine, and analyze the major obstacles for its clinical application, with an aim to provide some new clues for clinicians and researchers engaged in the clinical and basic research of HCC.

Functional versus non-functional intratumor heterogeneity in cancer

Next-generation sequencing data from human cancers are often difficult to interpret within the context of tumor evolution. We developed a mathematical model describing the accumulation of mutations under neutral evolutionary dynamics and showed that 323/904 cancers (∼30%) from multiple types were consistent with the neutral model of tumor evolution.

Allele surfing promotes microbial adaptation from standing variation

The coupling of ecology and evolution during range expansions enables mutations to establish at expanding range margins and reach high frequencies. This phenomenon, called allele surfing, is thought to have caused revolutions in the gene pool of many species, most evidently in microbial communities. It has remained unclear, however, under which conditions allele surfing promotes or hinders adaptation. Here, using microbial experiments and simulations, we show that, starting with standing adaptive variation, range expansions generate a larger increase in mean fitness than spatially uniform population expansions. The adaptation gain results from 'soft' selective sweeps emerging from surfing beneficial mutations. The rate of these surfing events is shown to sensitively depend on the strength of genetic drift, which varies among strains and environmental conditions. More generally, allele surfing promotes the rate of adaptation per biomass produced, which could help developing biofilms and other resource-limited populations to cope with environmental challenges.

Re-Evaluating Clonal Dominance in Cancer Evolution

Tumours are composed of genetically heterogeneous subclones which may diverge early during tumour growth. However, our strategies for treating and assessing outcome for patients are overwhelmingly based upon the classical linear paradigm for cancer evolution. Increasing numbers of studies are finding that minor subclones can determine clinical disease course, and that temporal and spatial heterogeneity needs to be considered in disease management. In this article we review evidence for cancer clonal heterogeneity, evaluating the importance of tumour subclones and their growth through both Darwinian and neutral evolution. Major shifts in current clinical practice and trial designs, aimed at understanding cancer evolution on a patient-by-patient basis, may be necessary to achieve more successful treatment of heterogeneous metastatic disease.

Genomics, microRNA, epigenetics, and proteomics for future diagnosis, treatment and monitoring response in upper GI cancers

One major objective for our evolving understanding in the treatment of cancers will be to address how a combination of diagnosis and treatment strategies can be used to integrate patient and tumor variables with an outcome-oriented approach. Such an approach, in a multimodal therapy setting, could identify those patients (1) who should undergo a defined treatment (personalized therapy) (2) in whom modifications of the multimodal therapy due to observed responses might lead to an improvement of the response and/or prognosis (individualized therapy), (3) who might not benefit from a particular toxic treatment regimen, and (4) who could be identified early on and thereby be spared the morbidity associated with such treatments. These strategies could lead in the direction of precision medicine and there is hope of integrating translational molecular data to improve cancer classifications. In order to achieve these goals, it is necessary to understand the key issues in different aspects of biotechnology to anticipate future directions of personalized and individualized diagnosis and multimodal treatment strategies. Providing an overview of translational data in cancers proved to be a challenge as different methods and techniques used to obtain molecular data are used and studies are based on different tumor entities with different tumor biology and prognoses as well as vastly different therapeutic approaches. The pros and cons of the available methodologies and the potential response data in genomics, microRNA, epigenetics and proteomics with a focus on upper gastrointestinal cancers are considered herein to allow for an understanding of where these technologies stand with respect to cancer diagnosis, prognosis and treatment.

Does the addition of drugs targeting the vascular endothelial growth factor pathway to first-line chemotherapy increase complete response? A meta-analysis of randomized clinical trials

Drugs targeting the vascular endothelial growth factor (VEGF) and its receptor (VEGFR) signaling (anti-VEGF/VEGFR drugs) are the most validated anti-angiogenic strategies for cancer treatment. Complete response (CR) is a rare event in cancer patients receiving chemotherapy. A meta-analysis was conducted to determine whether adding anti-VEGF/VEGFR drugs to chemotherapy can further increase the chance of CR in the first-line therapy. Relevant databases were systematically searched for the period 2000-2015. Eligible studies were selected according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. The incidence, relative risk (RR), and 95 % confidence intervals (CIs) were calculated using random-effects or fixed-effects models based on the heterogeneity of selected studies. A total of 12,453 patients from 28 randomized controlled trials were included. The overall incidence of CR in patients treated with anti-VEGF/VEGFR drugs plus chemotherapy was 1.5 % (95 % CI, 1.0-2.0 %) compared to 1.1 % (95 % CI, 0.7-1.4 %) in the chemotherapy-alone arm. Adding anti-VEGF/VEGFR drugs was associated with significant improvement of CR (RR, 1.52, 95 % CI, 1.18-1.95, P = 0.001). When stratified by drug type, adding VEGFR tyrosin kinase inhibitors (TKIs) did not increase the chance of CR (RR, 0.87, 95 % CI, 0.51-1.49; P = 0.614). The addition of bevacizumab with 7.5 mg/kg every 3 weeks, but not 15 mg/kg every 3 weeks, significantly improves the CR (7.5 mg, RR, 2.43, 95 % CI, 1.64-3.60, P = 0.000; 15 mg, RR, 1.07, 95 % CI, 0.63-1.81, P = 0.799). In subgroup analysis, a significant improvement of CR by the addition of anti-VEGF/VEGFR drugs was observed in patients with colorectal cancer (RR, 2.10, 95 % CI 1.21-3.63, P = 0.008), ovarian cancer (RR, 3.07; 95 % CI, 1.68-5.62, P = 0.000), and patients who are treated with platinum-based regimens (RR, 1.78, 95 % CI, 1.23-2.59, P = 0.002). Low-dose bevacizumab, rather than VEGFR TKIs or high-dose bevacizumab, can increase the chance of CR in patients receiving chemotherapy.