Journal of Molecular Medicine (2020) 98:161–177 https://doi.org/10.1007/s00109-020-01874-2 REVIEW Clinical implications of intratumor heterogeneity: challenges and opportunities Santiago Ramón y Cajal 1,2,3,4 & Marta Sesé 1,3 & Claudia Capdevila 1,5 & Trond Aasen 1,3 & Leticia De Mattos-Arruda 6 & Salvador J. Diaz-Cano 7 & Javier Hernández-Losa 1,2,3 & Josep Castellví 1,2,3 Received: 10 May 2019 / Revised: 5 November 2019 / Accepted: 7 January 2020 / Published online: 22 January 2020 # The Author(s) 2020 Abstract In this review, we highlight the role of intratumoral heterogeneity, focusing on the clinical and biological ramifications this phenomenon poses. Intratumoral heterogeneity arises through complex genetic, epigenetic, and protein modifications that drive phenotypic selection in response to environmental pressures. Functionally, heterogeneity provides tumors with significant adaptability. This ranges from mutual beneficial cooperation between cells, which nurture features such as growth and metastasis, to the narrow escape and survival of clonal cell populations that have adapted to thrive under specific conditions such as hypoxia or chemotherapy. These dynamic intercellular interplays are guided by a Darwinian selection landscape between clonal tumor cell populations and the tumor microenvironment. Understanding the involved drivers and functional consequences of such tumor heterogeneity is challenging but also promises to provide novel insight needed to confront the problem of therapeutic resistance in tumors. Keywords Intratumor heterogeneity . Liquid biopsy . Artificial intelligence . Antitumor therapeutics Background heterogeneity) refers to the differences found between tumors in different patients. Intratumor heterogeneity (also known as Malignant tumors have highly diverse phenotypic and molec- intralesion heterogeneity) refers to distinct tumor cell popula- ular characteristics both at the intertumor and intratumor tions (with different molecular and phenotypical profiles) levels [1]. Intertumor heterogeneity (also known as interlesion within the same tumor specimen [1]. * Santiago Ramón y Cajal 1 Translational Molecular Pathology, Vall d’Hebron Institute of sramon@vhebron.net Research (VHIR), Universitat Autònoma de Barcelona, Passeig Vall d’Hebron 119-129, 08035 Barcelona, Spain Marta Sesé 2 marta.sese@vhir.org Pathology Department, Vall d’Hebron Hospital, Passeig Vall d’Hebron 119-129, 08035 Barcelona, Spain Claudia Capdevila 3 cc4107@cumc.columbia.edu Spanish Biomedical Research Network Centre in Oncology (CIBERONC), Barcelona, Spain Trond Aasen 4 trond.aasen@vhir.org Department of Pathology, Vall d’Hebron University Hospital, Autonomous University of Barcelona, Pg. Vall d’Hebron, 119-129, Leticia De Mattos-Arruda 08035 Barcelona, Spain ldemattos@vhio.net 5 Salvador J. Diaz-Cano Department of Genetics and Development, Columbia University sdiaz-cano@nhs.net Medical Center, New York, NY 10032, USA 6 Javier Hernández-Losa Vall d’Hebron Institute of Oncology, Vall d’Hebron University jahernan@vhebron.net Hospital, c/Natzaret, 115-117, 08035 Barcelona, Spain 7 Josep Castellví Department of Histopathology, King’s College Hospital and King’s jocastellvi@vhebron.net Health Partners, London, UK 162 J Mol Med (2020) 98:161–177 Cancer is typically defined as a genetic disease driven Phenotypic heterogeneity: the particular case by oncogenic mutations. In a similar gene-centric view, of morphologic heterogeneity intratumoral heterogeneity has traditionally been attribut- ed to genetic diversity within cancer cell populations. Every tumor is unique as a result of its interactions with the host However, recent evidence suggests that a tumor is hetero- and the genetic and epigenetic variability. This results in impor- geneous in almost every discernible phenotypic trait as tant differences between tumors from different individuals, the result of not only genetic influences but also non- named as intertumoral heterogeneity. At this level, we recog- genetic sources of variability [2]. These non-genetic influ- nize more than 250 types of tumors with distinctive clinical- ences shape the phenotypic states of cancer cells at the pathological characteristics and that show most of them also proteome level. These factors are the primary determi- peculiar pathological characteristics. In most of these tumor nants of the identity of most cell types in healthy tissue, types, dozens or hundreds of pathological variants are observed. given that most cells, while phenotypically different, In such variability of tumors and, therefore, of types of cancer, share the same genetic load. factors such as the location and the cell type are determinant. Tumor heterogeneity is associated with poor prognosis Neoplastic lesions are diagnosed primarily based on patho- and outcome [3–6]. It is thought that intratumor heteroge- logical examination, both gross and microscopic, but the infor- neity is one of the leading determinants of therapeutic mation obtained may not always be conclusive for a diagnosis resistance and treatment failure and one of the main rea- of malignancy. In this context, intratumoral heterogeneity sons for poor overall survival in cancer patients with met- poses an unresolved problem. Most carcinomas, sarcomas, astatic disease [1, 7]. Tumors are composed of mosaics of and astrocytomas display extensive intratumoral morphologi- cancer cells with different characteristics and varying sen- cal variability that, if not taken into account, can lead to an sitivities to anticancer therapies. Tumor heterogeneity has inaccurate or even incorrect diagnosis. For example, analysis differential layers of complexity. Because cancer is a het- of complete specimens of lung adenocarcinoma often reveals erogeneous dynamic target, individual patients, lesions, more than one morphological pattern (acinar, solid, lipid, pap- and cell populations should be thoroughly characterized illary, micropapillary, mucinous, or pleomorphic) (Fig. 1) [15], at varying times. Tumor heterogeneity has presented a and accurate assessment is critical for the right diagnosis and considerable challenge to matching patients with the right prognosis. Therefore, diagnosis based on morphology requires treatment at the right time; therefore, it poses a challenge extensive areas of the tumor to be examined to ensure an ob- to accomplish the goals of precision medicine [8, 9]. jective, genuinely representative snapshot of the heterogeneity It has been shown that most of the targets considered as within the tumor as a whole. State-of-the-art digital image ac- druggable and for which Food and Drug Administration quisition and quantification algorithms, which integrate bio- (FDA)-approved therapeutics are available are not physical parameters to capture the spatial variation in tumor expressed in a uniform or homogeneous manner in tumor architectures, are likely to play an essential role in this [16]. tissue. Examples include the variable threshold of positiv- In many malignant tumors, it is common to find well- ity for the expression of the HER2 receptor in gastric differentiated areas adjacent to poorly or moderately differen- adenocarcinoma [10], progesterone and estrogen receptors tiated areas. Attempts are currently being made to quantify in breast tumors [11], the EML4-ALK translocation in these areas and grade tumors, generally according to the lung adenocarcinoma [12], B-RAF mutations in melano- least-differentiated area or the area with the highest degree ma, and the 1p/19q allelic loss in oligodendrogliomas [13, of cytologic malignancy [17, 18]. This intratumoral differen- 14]. This variability means that many patients undergo tiation is often patchy, and not well defined in molecular specific therapeutic regimens in situations where perhaps terms. It reproduces the development patterns and only 10% of the studied tumor cells are positive for the morphofunctional specialization present in the tissue where corresponding target. Thus, targeted therapies in heteroge- cancer originated and can lead to differences in the expression neous neoplasms lead to transient tumor regression and of some of the therapeutic targets and, therefore, the response subsequent selective outgrowth of existing resistant pop- to a specific treatment [10, 19–21]. One of the most charac- ulations, leading to recurrence in the long run. teristic examples is EGFR-driven lung adenocarcinoma in non In the following sections, we discuss intratumor hetero- small-cell lung cancer (NSCLC), as there have been cases of geneity and in particular how this is driven at the genetic resistance associated with conversion to small-cell lung cancer level and in relation to the microenvironment, and how (SCLC) phenotype after long-term treatment with EGFR ty- this translates into specific phenotypes at the functional rosine kinase inhibitors [22]. level. Finally, we discuss research approaches needed to Similarly, the observation of specific morphologic patterns advance our understanding of this complex biological in human tumors has made it possible to identify distinct ge- phenomenon and how this can lead to novel therapeutic netic changes. For example, characteristic chromosomal trans- approaches. locations have been identified in round cell desmoplastic J Mol Med (2020) 98:161–177 163 heterogeneity and the emergence of an eventual phenotype of resistance to many conventional treatments. Nevertheless, the puzzle and variability of cancer pa- thology are made tremendously complicated at the geno- mic level by the vast number of DNA changes, with thou- sands of known translocations and more than 1500 muta- tions, deletions, and amplifications reported to date [8, 26–29]. Also noteworthy is the complex world of microRNAs (miRNAs), of which there are thousands de- scribed (www.mirbase.org) and which can act like oncogenes or tumor suppressors (oncomiRs and tumor suppressor miRs) [30, 31], and the unknown role of long non-coding RNA (lncRNA), numbering as many as 60,000 loci in the human genome [32–34]. Importantly, there are many nonspecific genetic alterations in human tumors. For example, the ETV6-NTRK3 translocation can be detected in very different types of tumors such as infantile fibrosar- coma, cellular mesoblastic nephroma, and secretory carci- noma of the breast [35]. Chromosomal translocations such as those including the ALK gene are demonstrated in ana- plastic lymphoma, lung adenocarcinoma, and myofibroblastic tumors [36], and the translocation EWSR1-CREB1 in tumors as different as clear cell sarco- ma and angiomatoid fibrous histiocytoma [37]. BRAF mu- tations and translocations have been described in melanocytic nevi, malignant melanoma, colon adenocarci- Fig. 1 Lung cancer intratumoral heterogeneity at morphological and noma, glioblastoma and pilocytic astrocytoma [38, 39] and molecular levels. a Paraffin section of a lung tumor biopsy showing EGFR mutations, amplifications in lung adenocarcinoma three main morphological subtypes within the same tumor. b Molecular and brain tumors [40–42]. and biomarker analysis confirming heterogeneity in EGFR mutation and in the c transcriptional signature of these three subtypes Although some molecular alterations are recurrent in some tumors, not all the tumors of the same type, and similar mor- tumors, clear cell sarcoma, synovial sarcoma, and rhabdoid phology, show the same genetic profile. In fact, there is a huge tumors [23, 24]. Nevertheless, not all oncogenic changes are intertumoral heterogeneity between tumors with the same his- diagnostic determinants of a specific tumor type or give rise to tology in different individuals. This heterogeneity poses a the emergence of a specific morphologic pattern. Researchers problem when trying to standardize a therapy in a specific from the National Research Tomsk State University [25] have tumor type, requiring a personalized approach based on the shown that the morphological heterogeneity in invasive particular genetic alterations of the tumor, to improve the re- micropapillary carcinoma (IMPC) of the breast is not related sponse rates to the treatment. to the presence of specific chromosomal aberrations. This het- Moreover, initial studies have shown that a character- erogeneity responds to specific gene expression profiles, thus istic morphological pattern could be due to specific onco- pointing to the existence of other determinants of intratumor genic changes and that malignancy is dependent on the morphological heterogeneity and highlighting the importance immune response [43]. For example, primary cells with of context. Furthermore, some of the most common molecular RAS, NEU, mutated p53, MYC, and the viral gene E1A alterations have been associated with tumors whose morpho- oncogenes injected into athymic mice [43] form different logical characteristics are strikingly distinct. morphological patterns in melanoma [17]. In this sense, the first point to underline would be that the presence of different morphological patterns within the same tumor suggests the coexistence of various clones, each subject Molecular heterogeneity: a genomic substrate to specific genetic changes or different environmental for both tumor biology and evolution pressures that are not necessarily shared by clones present in other areas. During the 1990s, and after the discovery of oncogenes, it was Critically, this genetic variability is also thought to occur thought that specific genetic changes would account for tumor extensively within a tumor (Fig. 1). Therefore, the goal of 164 J Mol Med (2020) 98:161–177 preparing selective oncograms with chemotherapy drugs and that the most important aspect of tumor transformation other inhibitors has become complicated—almost impossi- and subsequent progression is the functional alteration of ble—following conventional strategies. Intratumor molecular at least ten major biochemical pathways. Given that phe- heterogeneity, and more specifically that occurring notypes generated by changes in genetic material are the intratumorally, is thought to account for the unsuccessful at- substrate of clonal development and selection and adapta- tempts of pharmacologic and radiologic cancer treatments. In tion to the microenvironment for each of these pathways fact, in cell lines, a correlation has previously been shown (e.g., insensitivity to apoptosis, self-sufficiency in cell between resistance to radiotherapy and cytotoxic drugs (e.g., proliferation, acquisition of so-called replicative immor- cisplatin, doxorubicin, and taxanes) and the expression of spe- tality), several significant genetic changes must occur cific oncogenes, principally, RAS, p53, c-MYC, and the ade- (Fig. 2) [62]. A representative example can be seen in noviral gene E1A [44–47]. signaling in the Ras/MAPK and PTEN/PI3K/AKT axes in lung adenocarcinomas, where specific mutations, am- Causes of molecular heterogeneity plifications, gene expression, and translocations in mem- brane receptors, as in other genes downstream, can enable There are a lot of redundant genetic alterations in cellular the tumor cell hallmark of uncontrolled proliferation [60, and biological pathways. These pathways can be grouped 63]. into specific “hallmarks” or basic principles that rational- This concept can be extrapolated to other significant path- ize tumor biology and that are altered in the vast majority ways, whose number is expected to grow to 15–20 in the of malignant neoplasms [48, 49]. Authors have postulated coming years [49]. Although “single hit pathways” are Fig. 2 Clonal cooperation and cellular consortium. a Darwinian model of clonal heterogeneity resulting in a consortium of clones, each with their characteristics and malignant features. b Cooperation between several clones to invade and metastasize J Mol Med (2020) 98:161–177 165 reported, disruptions of several pathways are necessary for a Many of these regional epigenetic differences are associat- cell to become malignant. Moreover, a similar biological ef- ed with an aberrant methylation pattern in specific promoters fect can be achieved by hitting a particular biochemical path- or other regulatory elements causing either gene activation or way at different points, often driven by the inherent genetic silencing [59, 72–77], which may also be predictive of the instability of a tumor, thus making cancer an extremely het- phylogenetic relationships between the different clones in tu- erogeneous (and redundant) disease at the molecular level [54, mors such as prostate cancer [78]. Therefore, we can deduce 64, 65] (Fig. 1). that genomic and epigenomic diversity are not mutually ex- clusive but can be explained by a unified evolutionary pro- cess, giving rise to more robust evolutionary models than Understanding the role of genomic instability clonal relationships inferred from genetic or epigenetic as an enabling characteristic of cancer datasets alone. A summary of main molecular events related with intratumoral heterogeneity is shown in Table 1. Despite claims in recent models that only three driver muta- tions are required for the development of various forms of Proteomic heterogeneity: going beyond the genome advanced cancer [66, 67], the number of molecular changes necessary to enable the emergence of a clinically relevant If the genetic diversity of constitutive alterations in DNA is tumor has, for some time, been assumed to be higher than enormous, then at the level of the proteome, this diversity previously thought, given standard mutation rates in any cell increases exponentially. Given that proteins are the final effec- type. This observation is supported by the large number of tors of all cellular pathways, along with small metabolites, it somatic mutations and epigenetic alterations found in most seems reasonable to think that the “ideal” targets for therapy tumor specimens (in the order of thousands), which points to are those protein factors that have the most stable expression the existence of molecular mechanisms that enable the muta- and activation in tumor cells. Therefore, it is essential to con- tional landscape of cancer cells to expand. Therefore, to ex- sider the proteomic heterogeneity of tumors. plain the substantial molecular variability inherent to malig- Even in tumors with constitutive genetic activation of nant tumors, there must be a background of replicative im- EGFR and HER2, the underlying pathways are not perma- mortality and genomic instability, which is associated with nently and homogeneously active in all cells [79–83]. At pres- abnormalities in DNA repair mechanisms and maintenance ent, approaches such as multispectral imaging of multiple pro- of genetic and chromosomal integrity (see review [68]). teins from a common signaling pathway allow the accessible, Chromosome instability (CIN) and microsatellite instability multiplexed elucidation of proteomic heterogeneity at the lev- (MSI) have been described as two alternative pathways to el of signal transduction [84]. Moreover, proteomic heteroge- cancer [53, 69, 70]. neity is not always a simple consequence of the heterogeneity found at the genetic level. In fact, it may be affected by the microenvironment and stress situations such as starving or Epigenetic heterogeneity hypoxia [62, 85]. Much attention has been paid to the role of the molecular Genetically identical cell populations can display remarkable pathways controlling RNA splicing [86, 87], the impact of the morphological diversity. One mechanism by which different expression of various protein isoforms, and how the local environmental stimuli drive such heterogeneity is by epige- environmental factors determine their levels [88]. However, netic modifications of the genome, which can persist over protein synthesis machinery and other translation regulators many cell divisions [61]. Therefore, within what we under- are also significantly modulated by local environmental con- stand as molecular heterogeneity, regional differences in epi- ditions. Control of protein synthesis (and preceded immedi- genetic status have been observed in different types of cancer ately by regulation of transcription) is considered one of the that can act, in much the same way as genetic alterations, as leading post-transcriptional mechanisms for control of gene drivers of the tumor process. For example, in colon adenocar- expression. This control is profoundly altered in cancer [56]. cinomas, a subtype harbors a relevant profile of epigenetic Alterations in the expression and activity of specific trans- alterations [64]; this is also seen in urological [65] and other lation factors and their inhibition by cellular stress conditions tumors [57, 58, 71]. Moreover, local hypoxia may induce the (e.g. hypoxia or lack of nutrients through various pathways) expression of histone demethylases and other epigenetic mod- are common to most human tumors (especially in advanced ifiers that subsequently modulate the expression of genes stages) [56]. The tumor takes control of translation by various linked to a specific phenotype (e.g., leading to epithelial to mechanisms to cover the demands associated with high pro- mesenchymal transition). Inflammatory cytokines are another liferation rates or to promote translation of specific messen- example, released by stromal or immune cells, which can alter gers that are favorable to tumor progression (survival, pro- DNA methylation and other epigenetic markers. angiogenic, invasion, and metastasis) (reviewed in [89]). 166 J Mol Med (2020) 98:161–177 Table 1 Table summarizing the aspects highlighted in this review correlating with the key molecular events related with intratumoral heterogeneity Key points Bibliography 1. Phenotypic heterogeneity 1.1. Hundreds of tumor types and thousands of subtypes Jamal-Hanjani et al. (2015) [1] 1.2. Different degree of cell differentiation (low-grade and Park et al. (2010) [18] Zhou et al. (2015) [17] high-grade tumor types) 1.3. Morphologic pattern association with genetic changes Sequist et al. (2011) [22] Nielsen et al. (2015) [24] Zack et al. (2013) [26] 1.4. Morphological heterogeneity in metastasis vs. primary Maddipati et al. (2015) [50] Hong et al. (2015) [51] tumor 2. Molecular heterogeneity 2.1. Intratumor heterogeneity and resistance to treatments Dagogo-Jack and Shaw (2018) [52] 2.2. Different types of molecular changes in coding genes JamalHanjani et al. (2017) [7] Sharma and Debinski 2.2.1 SNV (2018) [41] Karachaliou et al. (2015) [40] 2.2.2 Insertions and Deletions 2.2.3 Copy number variation 2.2.4 Rearrangements (i.e translocations) Skoulidis and Heymach (2019) [42] 2.3. Genomic Instability (CIN and MSI) Dagogo-Jack and Shaw (2018) [52] Andor et al. (2016) [53] 2.4. Molecular and biochemical redundancy in the several Logue and Morrison (2012) [54] pathways altered in malignant cells 2.5. Heterogeneity at genomic level is not always related with Ramon Y Cajal S et al. (2017) [55] Ramon Y Cajal S heterogeneity at proteomic level et al. (2018) [56] 3. Epigenetic heterogeneity 3.1 Different changes (histone modifications, DNA Kumar et al. (2018) [57] Dong et al. (2017) [58] Bhawal methylation) on the genome associated with gene et al. (2007) [59] silencing / gene activation 3.2. Deregulation of gene expression (overexpression or Agarwal R et al. (2017) [60] inhibition) 3.2.1 conding genes: mRNAs 3.2.2 non-coding RNAs: miRNAs, lncRNAs Raychaudhuri et al. (2012) [33] Eriksen et al. (2016) [31] Ramon y Cajal et al.(2019) [34] 3.3. Associated and associated with the microenvironment Assenov et al. (2018) [61] Yuan Y (2016) [62] Biological interactions among distinct tumor can secrete factors associated with poor prognosis (e.g., clones and the microenvironment: the stroma chemokines secreted by tumor-associated histiocytes, may have a significant impact on phenotypic macrophages or fibroblasts [the so-called cancer-associat- heterogeneity ed fibroblasts]). Signatures of released stromal factors have been thought to affect progression and tumor differ- Twenty years ago, we published the first histopathologic entiation, as well as invasiveness in adenocarcinomas evaluation of whether differentiation in squamous cell car- [92–95]. Therefore, the importance of the surrounding cinoma could be related to the components of the stroma stroma in intratumoral morphologic heterogeneity seems [90]. After previous studies by Dotto and Weinberg [91], evident, both regarding factors released by fibroblasts and who observed that normal fibroblasts could inhibit the factors released by inflammatory cells such as histiocytes growth of RAS-transformed keratinocytes in athymic and lymphocytes. Accordingly, tumor multifocality has mice, we observed that coinjection of normal fibroblasts been postulated as being associated with the underlying together with RAS-transformed keratinocytes induced be- stroma [96]. Morphological and genetic heterogeneity is nign or low-grade malignant squamous lesions with ex- the result of a multistep process of tumorigenesis that tensive areas of keratinization that were observable by leads to subclonal tumor cell populations with distinct morphology, by immunohistochemistry, and by electron traits, according to current paradigms [2, 5, 8, 21]. The microscopy [90]. Subsequently, it was concluded that this model we present herein incorporates complementary the- fibroblast-mediated differentiation was secondary to fac- ories of tumor evolution such as the big bang model or the tors such as signaling by transforming growth factor beta cancer stem cell hypothesis [97] (Fig. 2). It is increasingly (TGF-β). clear that understanding alterations within tumor cells is Years of research have shown that the peritumoral stro- only part of the picture, and we need to understand inter- ma in many malignant tumors play an important role and actions between tumors and their microenvironment to J Mol Med (2020) 98:161–177 167 account for multiple aspects of tumor progression and mutations and other genetic and chromosomal rearrangements therapeutic resilience [55, 98–100]. [21, 111], recover functionality of previously inactivated In this regard, the concept of cancer as a consortium of genes whose potential had been exploited in synthetic lethal clones and local factors has been proposed [101]. The concept interactions [112], activate cellular dedifferentiation and of clonal cooperation is based on a single clone being unable transdifferentiation programs [97], and even potentiate the de- to acquire all the necessary properties to be an invasive tumor, velopment of specific populations by non–cell-autonomous such that various clones must act synergistically and comple- mechanisms [113]. Thus, it is the adoption of both genetic mentarily to acquire the characteristics described by Hanahan and non-genetic subclonal changes that endows cancer with and Weinberg [49] and the proposed biochemical changes in enough phenotypical plasticity to adapt to microenvironmen- 10 or more cellular biochemical pathways. This cell coopera- tal pressures and successfully overcome the barriers posed by tion can be observed in cell clusters in metastatic develop- antitumoral therapy. Otherwise, dissecting tumor heterogene- ment: clusters of circulating tumor cells (CTCs) are associated ity involves emerging strategies such as multiregional se- with higher number of metastases than single circulating cells quencing, analysis of autopsy samples, single-cell sequencing, in models of breast cancer, pancreatic cancer, and melanoma and longitudinal analysis of liquid biopsy samples [52]. Rapid [50, 51, 102–106]. Moreover, our group made an effort to research autopsy of cancer patients can explain heterogeneity generate MDA-MB-231 breast cancer cell lines single clones processes including cancer evolution and acquired therapeutic and demonstrated that the clonal cooperation confers aggres- resistance [114–119]. siveness and tumor progression [106]. Numerous studies have shown how genetic variants The microenvironment plays a role in the adoption of phe- emerge after therapy and suggest that resistance and response notypes that may be clinically relevant and are contingent to therapy from that moment onwards are commonly deter- upon the implementation of metabolic gene expression pro- mined by genetic variants (see [8] and references therein). For grams and, as such, can be completely independent of the example, in colon adenocarcinoma, highly sensitive tech- acquisition of new drivers. One of the most explicit examples niques and application of anti-EGFR therapy have made it is the well-known role of the HIF family of transcription fac- possible to detect up to 70% of Ras mutations in blood in tors, which, under hypoxic conditions, trigger a set of adaptive series where the percentage diagnosed in the primary tumor transcriptional responses (tumor angiogenesis, cell was approximately 40–45% based on standard molecular metabolism, invasion, survival, therapeutic resistance, and techniques [120]. This is also true for non-small cell lung even differentiation and self-renewal) and seem to play a crit- cancer and EGFR mutations [40]. In this sense, therapy has ical role in tumor progression [107]. been considered both a source of variability and a selective filter, promoting the acquisition of new mutations and the selective proliferation of previously dormant, minority clones How to address tumor heterogeneity [21, 121]. in a clinical setting Given that the strategy of targeting cancer-initiating muta- tions has been applied with limited success [122], we believe Intratumor heterogeneity (and its genetic and non-genetic de- that better comprehension of the determinants of tumor het- terminants) is a dynamic phenomenon that is observed at mul- erogeneity is needed (especially in intratumor terms). tiple levels, and that follows a mainly Darwinian-type progres- Pathologists have the responsibility to make a correct and sion, although it is far more complicated than previously verifiable diagnosis, from their examination of tissue samples, thought. Unpredictable and often chaotic cellular reactions taking into consideration all the variables that underlie depending on oncogenic alterations and environmental factors intratumoral heterogeneity. Tumor progression assessment drive tumor progression and hold the key to interpreting tumor would ideally analyze at least two samples to compare the development. This concept is essential because the decisions biologic markers relevant for progression in both tumor cell made during a patient’s treatment are based on the study of clones and the microenvironment. The tumor clone markers biopsy specimens of the primary tumor by pathologists and include those involved in the tumorigenic expansion usually revolve around the oncogenic drivers known at the (proliferation) and invasion, the two leading forces driving time of diagnosis [2] (Fig. 3). Given the complex and constant progression. The tumor microenvironment analysis focuses development of tumor architecture, it is essential to under- the attention on the qualities that potentiate clonal expansion stand that molecular changes (both genetic and epigenetic) and invasion of tumor cells. In essence, tumor progression within the tumor itself evolve during disease progression analysis must concentrate on clonal heterogeneity and over- and metastasis [108]. Therefore, the biopsy of a primary tumor come the problems it presents. is not necessarily predictive of what happens in secondary One important aspect is how representative biopsies re- deposits [2]. In addition, chemotherapy and radiotherapy can flects the overall tumor histology and biology. Core biopsies trigger selection of resistant clones [2, 109, 110], induce new often only reflect a spatiotemporal snapshot of the whole 168 J Mol Med (2020) 98:161–177 Fig. 3 Cancer biology-driven personalized medicine. Schematic representation of the clinical workflow for lung cancer diagnosis, treatment, and follow- up tumor and are therefore unlikely to be fully informative about protein expression profiling (especially analyses including the clonal composition [123–129]. The size of the sample is next-generation sequencing (NGS) techniques) and be able another critical issue [130–132], and signal-to-noise ratios to include gene signatures that are characteristic of a different need to be balanced. One way to achieve this balance is isolate prognosis or clinical treatment [55, 70, 92, 94, 130, 135–137] by microdissection multiple relatively small regions of tumors (Fig. 3). The incorporation of NGS and the development of that more likely represent the balance of morphologically dis- new resources for the analysis of these big data, combining tinct units. The importance of such approach is highlighted by molecular and expression signatures, are becoming crucial for the observation of clustered populations within a tumor that diagnosis [13, 138]. The field of radiogenomics, which corre- differ in gene expression [133], as well as genetic composition lates genomic data with the radiological features of the tu- [134]. However, unless large numbers of samples are provided mors, must also be taken into account [139, 140]. While this for each tumor, this approach can easily fail to identify patches approach based on artificial intelligence may be interesting for of genetically distinct cells [130–132]. On the other hand, the differential diagnosis of radiological features, we under- larger samples, or pools of samples, lead to intermixing of stand that genomic information from a single sample is not small anatomically distinct units, which provides additional necessarily representative of the whole tumor and its hetero- challenges in relation to distinguish distinct functional hetero- geneity. While procuring multiple metastatic tumor samples geneity. Multiple solid biopsy samples should be taken based for genomic studies through NGS and development of patient- on data obtained through imaging and nuclear medicine, with derived xenografts or organoids, mechanistic insights gained the selection of the biopsied area relying increasingly on from research autopsy studies of cancer patients can help iden- criteria such as particular metabolic activity. It is also essential tify new targets for therapeutic intervention [114]. In collabo- to bear in mind that microenvironmental factors such as hyp- ration with Cambridge CRUK, our group has performed ex- oxia and inflammatory infiltrate can induce changes in the tensive multi-platform profiling of metastases in 10 warm au- protein expression of therapeutic targets and condition the topsies of patients with lethal multi-therapy-resistant breast response to antitumor agents. Therefore, we must select the cancers (DNA sequencing, RNA sequencing, the T cell recep- most representative areas for massive parallel sequencing and tor (TCR) sequencing, and immunohistochemistry (IHC)) of genomic and proteomic studies and report on their limitations. multiple individual metastases (range 5–36 metastases per The histopathological diagnosis should integrate molecular case, 182 individual metastases to 22 organ sites). This col- analysis (genome sequencing, transcriptome profiling) and lection allowed us to characterize the mutational and copy J Mol Med (2020) 98:161–177 169 number aberration (CNA) landscapes across the individual Every effort should be made to form multidisciplinary metastasis, to infer the clonal ancestries of metastases, to as- teams involving radiologists, nuclear medicine specialists, pa- sess the TME in each individual metastasis, to characterize the thologists, oncologists, systems biologists, molecular biolo- predicted neo-antigens, and to assess the TCR repertoires gists, and data scientists. Tumors must be analyzed at the across metastases, providing an unprecedented molecular genetic, molecular, and clinical-radiological level, with inte- characterization of lethal breast cancers that had been subject- gration and correlation of findings to ensure a holistic ed to multiple lines of systemic therapies [119]. approach. Finally, one of the most powerful techniques is the study of tumor heterogeneity at the cellular level. This approach, called (2) To overcome tumor heterogeneity, research should be single-cell sequencing, is based on the isolation of dozens of directed towards the search for central nodes, funnel fac- cells in different areas of the tumor, and the study of various tors, master regulator genes, and non-oncogene addic- Multi(omics) over them [141, 142]. For example, DNA se- tions [122, 147–149], in an attempt to confer therapeutic quencing after gene amplification can allow the study of mu- sensitivity. Regarding drug development in malignant tations, amplifications, deletions, and translocations in various tumors and current paradigms in cancer research, new areas of the tumor, thus characterizing the homogeneity of agents include those that target cancer-related vulnerabil- these genetic alterations [141, 143, 144]. Expression studies ities in receptor tyrosine kinases and intracellular signal- are also done, both at the RNA level (RNA-Seq) and epige- ing pathways, epigenetics, metabolism, and nuclear- netics with methyloma sequencing [143, 145, 146]. These cytoplasmic transport, among others. The study of the studies, nowadays, can be expensive and tedious in time as tumor immune microenvironment appears quite promis- well as in their interpretation, but are already showing results ing and includes treatment with immune checkpoint an- of high clinical interest; for example, the identification of het- tibodies, with programmed death 1 (PD-1 and PDL-1) erogeneity of mutations of the PIK3CA gene in breast cancer targeted agents, and novel immunotherapies. It is likely with HER2 amplification where the authors describe that that combinations will be needed for most subtypes. PIK13A and HER2 are not always present in the same cells Recent studies in solid cancers have highlighted the pres- and that chemotherapy selected the cells with mutant PIK3CA ence and relevance of immune heterogeneity and that [111]. It is an example of the importance of studying intratumor heterogeneity may also influence the anti- intratumoral molecular heterogeneity and where single-cell tumor immune responses [150–152]. sequencing technology can be decisive. Several studies [153–160] have shown that the ex- pression of factors such as 4EBP1 and EIF4E is diffuse in most solid tumors and glioblastomas and is associated How do we envision cancer research with lower survival and poorer prognosis. We proposed and treatment in the coming years? the concept of funnel factors [80], that is, factors that channel crucial information on tumor progression inde- (1) To assess intratumoral heterogeneity of tumors efficient- pendently of the level at which a specific oncogenic al- ly, it is essential to systematically integrate molecular teration occurs. These factors, which play a significant patterns, protein expression, and morphology into the role in the control of protein synthesis, could be sensitive fuller context of all clinical and pathological information tumor targets in a large number of malignant tumors [79, available (Fig. 3). We have proposed the term 83, 161, 162]. tissunomics, whereby a diagnosis is individually Complex models that implement combinatorial therapy assessed based upon a combined picture derived from are likely to be particularly beneficial in tumors with a the clinical, pathological, molecular, and protein expres- high degree of tumor heterogeneity. In this broad context, sion data of the tumor and its surrounding microenviron- evolutionary clues and new findings on interclonal rela- ment [55]. Importantly, molecular diagnosis based on tionships should also be taken into account [81, 101, 113, small samples and genetic alterations can lead to a false 163]. The identification of factors involved in this inter- negative diagnosis or treatment due to genetic and epi- play between malignant clones, which mediate tumor genetic changes present in a small subset of tumor cells. growth and metastasis, may be one promising approach In addition, tumor type and location has been shown to in the understanding of cancer [101]. Therefore, studies underlie unpredictable treatment responses targeting the carried out from the perspective of systems biology [149], same molecular pathway, such as the tumor response in tailored towards the identification of hubs or other central melanomas vs. colon carcinomas with BRAF mutations. factors in this complicated tangle of biochemical networks More conclusive data from basket trials and umbrella responsible for maintaining the tumorigenic state, will be trials are needed [55]. fundamental in the identification of addictions and 170 J Mol Med (2020) 98:161–177 vulnerabilities in cancer that would otherwise be difficult (4) Artificial intelligence. Intratumor heterogeneity is one of to imagine [147, 164]. the main reasons for the lack of diagnostic reproducibil- ity between pathologists given the complexity of the mi- (3) Liquid biopsies. Difficulties in obtaining tumor tissue croscopic interpretation of certain tumors. Furthermore, using invasive surgical procedures have led to the devel- many biomarkers do not have an established interpreta- opment of liquid biopsies for several cancer types tion algorithm. It is critical to improve existing algo- [165–184]. They comprise tumor-derived nucleic acids rithms for the quantification of immunohistochemical (e.g., circulating cell-free tumor DNA [ctDNA], and other in situ biomarkers. The development of artifi- microRNA), circulating tumor cells (CTCs), and cial intelligence algorithms with automatic learning tumor-derived extracellular vesicles that accumulate in (“deep learning”) is already shaping the field. Deep the blood, cerebrospinal fluid (CSF), urine, saliva, and learning methodology, with the generation of thousands other fluids [165, 178, 185–191]. One advantage of liq- of clinical-pathological diagnostic cases, can promote uid biopsies is that it significantly reduces the problem of the development of algorithms based on this methodolo- spatial heterogeneity. Several studies, comparing blood gy that could represent a breakthrough in the pathologi- and tissue biopsies, have confirmed that this approach cal diagnosis As an example, Google released has high specificity, although variable sensitivity is re- TensorFlow, an algorithmic development framework ported. Another important advantage (although under for distributed computing, to the general scientific and certain situations it may be a disadvantage) is that it tends technical community. This open-source machine learn- to reflect an aggregate of the output (ctDNA/CTC etc.) ing tool is free for any qualified scientist and is special- potentially from both primary and various metastatic ized in cognitive computing. sites. Such complex tumor heterogeneity cannot be eval- uated by a single core tumor needle biopsy [192]. With this approach, software is being developed by many startups and educational institutions as well as big companies However, the most clinically advanced approach is ctDNA such as Google, Phillips and Leica Microsoft. Algorithm- from plasma which closely matches the gene profile of tumor related applications for primary diagnosis, intraoperative di- tissue biopsies. Plasma ctDNA provides tumor-derived mate- agnosis, training, quantification of immunohistochemistry, or rial to identify actionable genomic alterations, monitor treat- diagnostic consultation are likely to progress significantly ment responses, predict progression of the tumor before clin- over the next few years. Notably, there have been several ical or radiological confirmation, and can identify mecha- claims that the accuracy and reliability of diagnoses based nisms of resistance also during therapy [173, 174, 176, 193, on neural network systems is very high [197, 198]. 194]. For a comprehensive review, see [195]. Examples have been published for skin cancer (both melano- Prospective clinical studies using liquid biopsies have ma and squamous cell carcinoma), lung adenocarcinoma, gli- characterized and monitored over time the genomic al- oma, gastric carcinoma, and others [135, 199–202]. terations of patients [40, 174]. Recently, the TRACERx Moreover, the deep learning tumor prediction heat map can consortium [7, 196] investigated tumor heterogeneity be quite complementary to pathologists’ “workflow.” An al- and evolution in early-stage NSCLC and showed the gorithm can detect, for example, metastatic carcinoma in prognostic value of copy-number heterogeneity assess- lymph nodes, or tumor budding in the colon or cervix, and ment in tumor biopsies and circulating tumor DNA de- help to recognize histologic patterns associated with higher tection in plasma. However, these liquid biopsy results malignant grades in gliomas [203], and moreover, can score reflect a kind of summary of tumor burden, regardless the degree of malignancy in tumors such a prostate adenocar- of the origin of the tumor cells (from primary or meta- cinomas where quantification of the histological patterns are static deposits), and require some degree of by-pass of underway [197, 198]. microanatomical boundaries (vascular basement mem- These algorithms are likely to help pathologists in reaching brane and stromal invasion) by either active tumor in- a faster, more accurate diagnosis and significantly reduce the vasion or passive external damage (e.g., ischemic or pathologist-dependent discordance in histopathological inflammatory). In this context, some caution should be diagnosis. taken for the evaluation of early epithelial neoplasms. The role of subclonal driver events in response to ther- (5) As a final reflection, we firmly believe that research strat- apy and disease recurrence and progression remains to be egies should be optimized. At present, most research determined. The use of liquid biopsies may pave the way teams are small, self-managed groups. Consequently, re- for a more detailed, real-time patient-tracking approach search is slow, and financial and human resources are not allowing the modification of therapeutic strategies optimized. We must establish more rational and ambi- throughout the disease. tious organizational models and strategies, with real J Mol Med (2020) 98:161–177 171 networks and professional, well-trained teams. As et al (2014) Single-cell RNA-seq highlights intratumoral hetero- geneity in primary glioblastoma. Science 344(6190):1396–1401 Horning recently said [204], “science and technology 7. Jamal-Hanjani M, Wilson GA, McGranahan N, Birkbak NJ, are at an inflection point with convergence—the integra- Watkins TBK, Veeriah S, Shafi S, Johnson DH, Mitter R, tion of life sciences, physical sciences, mathematics, en- Rosenthal R et al (2017) Tracking the evolution of non-small- gineering, and information technology—poised to make cell lung cancer. N Engl J Med 376(22):2109–2121 8. McGranahan N, Favero F, de Bruin EC, Birkbak NJ, Szallasi Z, significant progress”. We must look forward and not for- Swanton C (2015) Clonal status of actionable driver events and get that our primary objective is to cure cancer or at least the timing of mutational processes in cancer evolution. Sci Transl make it a chronic disease. Such a social commitment Med 7(283):283ra254 requires us to search for all possible methods of cooper- 9. McGranahan N, Swanton C (2017) Clonal heterogeneity and tu- mor evolution: past, present, and the future. Cell 168(4):613–628 ation among those involved in the diagnosis and treat- 10. Gomez-Martin C, Concha A, Corominas JM, Garcia-Caballero T, ment of cancer. Garcia-Garcia E, Iglesias M, Lopez JA, Ramon y Cajal S, Rojo F, Palacios J et al (2011) Consensus of the Spanish Society of Funding information SRYC received support from Fondo de Medical Oncology (SEOM) and Spanish Society of Pathology Investigaciones Sanitarias (PI14/01320 and PI17/02247), Redes (SEAP) for HER2 testing in gastric carcinoma. Clin Transl Temáticas de Investigación Cooperativa en Salud (RD12/0036/0057), Oncol 13(9):636–651 CIBERONC (CB16/12/00363), and Generalitat de Catalunya 11. Hammond ME, Hayes DF, Dowsett M, Allred DC, Hagerty KL, (AGAUR, 2017 SGR 1799 and 2014 SGR 1131). TA received support Badve S, Fitzgibbons PL, Francis G, Goldstein NS, Hayes M et al from Instituto de Salud Carlos III (grants PI16/00772 and CPII16/00042), (2010) American Society of Clinical Oncology/College Of co-financed by the European Regional Development Fund (ERDF). American Pathologists guideline recommendations for immuno- histochemical testing of estrogen and progesterone receptors in breast cancer. J Clin Oncol 28(16):2784–2795 Compliance with ethical standards 12. Abe H, Kawahara A, Azuma K, Taira T, Takase Y, Fukumitsu C, Murata K, Yamaguchi T, Akiba J, Ishii H et al (2015) Heterogeneity Conflict of interest The authors declare that they have no conflict of of anaplastic lymphoma kinase gene rearrangement in non-small- interest. cell lung carcinomas: a comparative study between small biopsy and excision samples. J Thorac Oncol 10(5):800–805 Open Access This article is licensed under a Creative Commons 13. Eckel-Passow JE, Lachance DH, Molinaro AM, Walsh KM, Attribution 4.0 International License, which permits use, sharing, adap- Decker PA, Sicotte H, Pekmezci M, Rice T, Kosel ML, Smirnov tation, distribution and reproduction in any medium or format, as long as IV et al (2015) Glioma groups based on 1p/19q, IDH, and TERT you give appropriate credit to the original author(s) and the source, pro- promoter mutations in tumors. N Engl J Med 372(26):2499–2508 vide a link to the Creative Commons licence, and indicate if changes were 14. Fiskus W, Mitsiades N (2016) B-Raf inhibition in the clinic: pres- made. The images or other third party material in this article are included ent and future. Annu Rev Med 67:29–43 in the article's Creative Commons licence, unless indicated otherwise in a 15. Travis WD, Brambilla E, Noguchi M, Nicholson AG, Geisinger credit line to the material. If material is not included in the article's K, Yatabe Y, Ishikawa Y, Wistuba I, Flieder DB, Franklin W et al Creative Commons licence and your intended use is not permitted by (2013) Diagnosis of lung cancer in small biopsies and cytology: statutory regulation or exceeds the permitted use, you will need to obtain implications of the 2011 International Association for the Study of permission directly from the copyright holder. To view a copy of this Lung Cancer/American Thoracic Society/European Respiratory licence, visit http://creativecommons.org/licenses/by/4.0/. Society classification. Arch Pathol Lab Med 137(5):668–684 16. O'Connor JP, Rose CJ, Waterton JC, Carano RA, Parker GJ, Jackson A (2015) Imaging intratumor heterogeneity: role in ther- apy response, resistance, and clinical outcome. Clin Cancer Res 21(2):249–257 References 17. Zhou M, Li J, Cheng L, Egevad L, Deng FM, Kunju LP, Magi- Galluzzi C, Melamed J, Mehra R, Mendrinos S et al (2015) Diagnosis of “poorly formed glands” Gleason pattern 4 prostatic 1. Jamal-Hanjani M, Quezada SA, Larkin J, Swanton C (2015) adenocarcinoma on needle biopsy: an interobserver reproducibil- Translational implications of tumor heterogeneity. Clin Cancer ity study among urologic pathologists with recommendations. Am Res 21(6):1258–1266 J Surg Pathol 39(10):1331–1339 2. Marusyk A, Almendro V, Polyak K (2012) Intra-tumour heteroge- 18. Park SY, Gonen M, Kim HJ, Michor F, Polyak K (2010) Cellular neity: a looking glass for cancer? Nat Rev Cancer 12(5):323–334 and genetic diversity in the progression of in situ human breast 3. Mroz EA, Rocco JW (2013) MATH, a novel measure of intratumor carcinomas to an invasive phenotype. J Clin Invest 120(2):636–644 genetic heterogeneity, is high in poor-outcome classes of head and 19. Denisov EV, Litviakov NV, Zavyalova MV, Perelmuter VM, neck squamous cell carcinoma. Oral Oncol 49(3):211–215 Vtorushin SV, Tsyganov MM, Gerashchenko TS, Garbukov EY, 4. Landau DA, Carter SL, Stojanov P, McKenna A, Stevenson K, Slonimskaya EM, Cherdyntseva NV (2014) Intratumoral morpho- Lawrence MS, Sougnez C, Stewart C, Sivachenko A, Wang L et al logical heterogeneity of breast cancer: neoadjuvant chemotherapy (2013) Evolution and impact of subclonal mutations in chronic efficiency and multidrug resistance gene expression. Sci Rep 4:4709 lymphocytic leukemia. Cell 152(4):714–726 20. Seol H, Lee HJ, Choi Y, Lee HE, Kim YJ, Kim JH, Kang E, Kim 5. Zhang J, Fujimoto J, Wedge DC, Song X, Seth S, Chow CW, Cao SW, Park SY (2012) Intratumoral heterogeneity of HER2 gene Y, Gumbs C, Gold KA, Kalhor N et al (2014) Intratumor hetero- amplification in breast cancer: its clinicopathological significance. geneity in localized lung adenocarcinomas delineated by Mod Pathol 25(7):938–948 multiregion sequencing. Science 346(6206):256–259 21. McGranahan N, Swanton C (2015) Biological and therapeutic 6. Patel AP, Tirosh I, Trombetta JJ, Shalek AK, Gillespie SM, impact of intratumor heterogeneity in cancer evolution. Cancer Wakimoto H, Cahill DP, Nahed BV, Curry WT, Martuza RL Cell 27(1):15–26 172 J Mol Med (2020) 98:161–177 22. Sequist LV, Waltman BA, Dias-Santagata D, Digumarthy S, Turke in circulating free DNA with survival in the EURTAC trial. JAMA AB, Fidias P, Bergethon K, Shaw AT, Gettinger S, Cosper AK et al Oncol 1(2):149–157 (2011) Genotypic and histological evolution of lung cancers acquir- 41. Sharma P, Debinski W (2018) Receptor-targeted glial brain tumor ing resistance to EGFR inhibitors. Sci Transl Med 3(75):75ra26 therapies. Int J Mol Sci 19(11) 23. Chang F (2006) Desmoplastic small round cell tumors: cytologic, 42. Skoulidis F, Heymach JV (2019) Co-occurring genomic alter- histologic, and immunohistochemical features. Arch Pathol Lab ations in non-small-cell lung cancer biology and therapy. Nat Med 130(5):728–732 Rev Cancer 19(9):495–509 24. Nielsen TO, Poulin NM, Ladanyi M (2015) Synovial sarcoma: 43. Ramon y Cajal S, Suster S, Halaban R, Filvaroff E, Dotto GP recent discoveries as a roadmap to new avenues for therapy. (1991) Induction of different morphologic features of malignant Cancer Discov 5(2):124–134 melanoma and pigmented lesions after transformation of murine 25. Denisov EV, Skryabin NA, Vasilyev SA, Gerashchenko TS, melanocytes with bFGF-cDNA and H-ras, myc, neu, and E1a Lebedev IN, Zavyalova MV, Cherdyntseva NV, Perelmuter VM oncogenes. Am J Pathol 138(2):349–358 (2015) Relationship between morphological and cytogenetic het- 44. Sanchez-Prieto R, Lleonart M, Ramon y Cajal S (1995) Lack of erogeneity in invasive micropapillary carcinoma of the breast: a correlation between p53 protein level and sensitivity of DNA- report of one case. J Clin Pathol 68(9):758–762 damaging agents in keratinocytes carrying adenovirus E1a mu- 26. Zack TI, Schumacher SE, Carter SL, Cherniack AD, Saksena G, tants. Oncogene 11(4):675–682 Tabak B, Lawrence MS, Zhsng CZ, Wala J, Mermel CH et al 45. Sanchez-Prieto R, Vargas JA, Carnero A, Marchetti E, Romero J, (2013) Pan-cancer patterns of somatic copy number alteration. Durantez A, Lacal JC, Ramon y Cajal S (1995) Modulation of Nat Genet 45(10):1134–1140 cellular chemoresistance in keratinocytes by activation of different 27. Mateo L, Guitart-Pla O, Pons C, Duran-Frigola M, Mosca R, Aloy oncogenes. Int J Cancer 60(2):235–243 P (2017) A PanorOmic view of personal cancer genomes. Nucleic 46. Sanchez-Prieto R, Quintanilla M, Cano A, Leonart ML, Martin P, Acids Res 45(W1):W195–W200 Anaya A, Ramon y Cajal S (1996) Carcinoma cell lines become 28. Roychowdhury S, Chinnaiyan AM (2016) Translating cancer ge- sensitive to DNA-damaging agents by the expression of the ade- nomes and transcriptomes for precision oncology. CA Cancer J novirus E1A gene. Oncogene 13(5):1083–1092 Clin 66(1):75–88 47. Duque PM, Alonso C, Sanchez-Prieto R, Quintanilla M, Ramon 29. Weinstein JN, Collisson EA, Mills GB, Shaw KR, Ozenberger S, Ramon y Cajal S (1998) Antitumoral effect of E1B defective BA, Ellrott K, Shmulevich I, Sander C, Stuart JM (2013) The adenoviruses in human malignant cells. Gene Ther 5(2):286–287 cancer genome atlas pan-cancer analysis project. Nat Genet 48. Hanahan D, Weinberg RA (2000) The hallmarks of cancer. Cell 45(10):1113–1120 100(1):57–70 30. Lin S, Gregory RI (2015) MicroRNA biogenesis pathways in 49. Hanahan D, Weinberg RA (2011) Hallmarks of cancer: the next cancer. Nat Rev Cancer 15(6):321–333 generation. Cell 144(5):646–674 31. Eriksen AH, Andersen RF, Nielsen BS, Sorensen FB, Appelt AL, 50. Maddipati R, Stanger BZ (2015) Pancreatic cancer metastases har- Jakobsen A, Hansen TF (2016) Intratumoral heterogeneity of bor evidence of polyclonality. Cancer Discov 5(10):1086–1097 microRNA expression in rectal cancer. PLoS One 11(6):e0156919 51. Hong MK, Macintyre G, Wedge DC, Van Loo P, Patel K, Lunke S, 32. Iyer MK, Niknafs YS, Malik R, Singhal U, Sahu A, Hosono Y, Alexandrov LB, Sloggett C, Cmero M, Marass F et al (2015) Barrette TR, Prensner JR, Evans JR, Zhao S et al (2015) The Tracking the origins and drivers of subclonal metastatic expansion landscape of long noncoding RNAs in the human transcriptome. in prostate cancer. Nat Commun 6:6605 Nat Genet 47(3):199–208 52. Dagogo-Jack I, Shaw AT (2018) Tumour heterogeneity and resis- 33. Raychaudhuri M, Schuster T, Buchner T, Malinowsky K, Bronger tance to cancer therapies. Nat Rev Clin Oncol 15(2):81–94 H, Schwarz-Boeger U, Hofler H, Avril S (2012) Intratumoral het- 53. Andor N, Graham TA, Jansen M, Xia LC, Aktipis CA, Petritsch C, erogeneity of microRNA expression in breast cancer. J Mol Diagn Ji HP, Maley CC (2016) Pan-cancer analysis of the extent and 14(4):376–384 consequences of intratumor heterogeneity. Nat Med 22(1):105– 34. Ramon YCS, Segura MF, Hummer S (2019) Interplay between 113 ncRNAs and cellular communication: a proposal for understand- 54. Logue JS, Morrison DK (2012) Complexity in the signaling net- ing cell-specific signaling pathways. Front Genet 10:281 work: insights from the use of targeted inhibitors in cancer therapy. 35. Argani P, Fritsch M, Kadkol SS, Schuster A, Beckwith JB, Genes Dev 26(7):641–650 Perlman EJ (2000) Detection of the ETV6-NTRK3 chimeric 55. Ramon YCS, Hummer S, Peg V, Matias Guiu X, De Torres I, RNA of infantile fibrosarcoma/cellular congenital mesoblastic Castellvi J, Martinez-Saez E, Hernandez-Losa J (2019) nephroma in paraffin-embedded tissue: application to challenging Integrating clinical, molecular, proteomic and histopathological pediatric renal stromal tumors. Mod Pathol 13(1):29–36 data within the tissue context: Tissunomics. Histopathology. 36. Taylor BS, Barretina J, Maki RG, Antonescu CR, Singer S, https://doi.org/10.1111/his.13828 Ladanyi M (2011) Advances in sarcoma genomics and new ther- 56. Ramon YCS, Castellvi J, Hummer S, Peg V, Pelletier J, Sonenberg apeutic targets. Nat Rev Cancer 11(8):541–557 N (2018) Beyond molecular tumor heterogeneity: protein synthe- 37. Cantile M, Marra L, Franco R, Ascierto P, Liguori G, De Chiara A, sis takes control. Oncogene 37(19):2490–2501 Botti G (2013) Molecular detection and targeting of EWSR1 fu- 57. Kumar R, Liu APY, Orr BA, Northcott PA, Robinson GW (2018) sion transcripts in soft tissue tumors. Med Oncol 30(1):412 Advances in the classification of pediatric brain tumors through 38. Hall RD, Kudchadkar RR (2014) BRAF mutations: signaling, DNA methylation profiling: from research tool to frontline diag- epidemiology, and clinical experience in multiple malignancies. nostic. Cancer 124(21):4168–4180 Cancer Control 21(3):221–230 58. Dong N, Shi L, Wang DC, Chen C, Wang X (2017) Role of 39. Penman CL, Faulkner C, Lowis SP, Kurian KM (2015) Current epigenetics in lung cancer heterogeneity and clinical implication. understanding of BRAF alterations in diagnosis, prognosis, and ther- Semin Cell Dev Biol 64:18–25 apeutic targeting in pediatric low-grade gliomas. Front Oncol 5:54 59. Bhawal UK, Tsukinoki K, Sasahira T, Sato F, Mori Y, Muto N, 40. Karachaliou N, Mayo-delas Casas C, Queralt C, de Aguirre I, Sugiyama M, Kuniyasu H (2007) Methylation and intratumoural Melloni B, Cardenal F, Garcia-Gomez R, Massuti B, Sanchez heterogeneity of 14-3-3 sigma in oral cancer. Oncol Rep 18(4): JM, Porta R et al (2015) Association of EGFR L858R mutation 817–824 J Mol Med (2020) 98:161–177 173 60. Agarwal R, Narayan J, Bhattacharyya A, Saraswat M, Tomar AK 79. Castellvi J, Garcia A, Rojo F, Ruiz-Marcellan C, Gil A, Baselga J, (2017) Gene expression profiling, pathway analysis and subtype Ramon y Cajal S (2006) Phosphorylated 4E binding protein 1: a classification reveal molecular heterogeneity in hepatocellular car- hallmark of cell signaling that correlates with survival in ovarian cinoma and suggest subtype specific therapeutic targets. Cancer cancer. Cancer 107(8):1801–1811 Genet 216-217:37–51 80. Armengol G, Rojo F, Castellvi J, Iglesias C, Cuatrecasas M, Pons 61. Assenov Y, Brocks D, Gerhauser C (2018) Intratumor heteroge- B, Baselga J, Ramon y Cajal S (2007) 4E-binding protein 1: a key neity in epigenetic patterns. Semin Cancer Biol 51:12–21 molecular “funnel factor” in human cancer with clinical implica- 62. Yuan Y, Jiang YC, Sun CK, Chen QM (2016) Role of the tumor tions. Cancer Res 67(16):7551–7555 microenvironment in tumor progression and the clinical applica- 81. Pons B, Peg V, Vazquez-Sanchez MA, Lopez-Vicente L, tions (review). Oncol Rep 35(5):2499–2515 Argelaguet E, Coch L, Martinez A, Hernandez-Losa J, 63. Devarakonda S, Morgensztern D, Govindan R (2015) Genomic Armengol G, Ramon YCS (2011) The effect of p-4E-BP1 and alterations in lung adenocarcinoma. Lancet Oncol 16(7):e342– p-eIF4E on cell proliferation in a breast cancer model. Int J e351 Oncol 39(5):1337–1345 64. Hinoue T, Weisenberger DJ, Lange CP, Shen H, Byun HM, Van 82. Serrano C, Romagosa C, Hernandez-Losa J, Simonetti S, Valverde Den Berg D, Malik S, Pan F, Noushmehr H, van Dijk CM et al C, Moline T, Somoza R, Perez M, Velez R, Verges R et al (2016) (2012) Genome-scale analysis of aberrant DNA methylation in RAS/MAPK pathway hyperactivation determines poor prognosis colorectal cancer. Genome Res 22(2):271–282 in undifferentiated pleomorphic sarcomas. Cancer 122(1):99–107 65. Lobo J, Barros-Silva D, Henrique R, Jeronimo C (2018) The 83. Martinez A, Sese M, Losa JH, Robichaud N, Sonenberg N, Aasen emerging role of Epitranscriptomics in Cancer: focus on urologi- T, Ramon YCS (2015) Phosphorylation of eIF4E confers resis- cal tumors. Genes (Basel) 9(11) tance to cellular stress and DNA-damaging agents through an 66. Tomasetti C, Marchionni L, Nowak MA, Parmigiani G, interaction with 4E-T: a rationale for novel therapeutic ap- Vogelstein B (2015) Only three driver gene mutations are required proaches. PLoS One 10(4):e0123352 for the development of lung and colorectal cancers. Proc Natl 84. Gerdes MJ, Sood A, Sevinsky C, Pris AD, Zavodszky MI, Ginty F Acad Sci U S A 112(1):118–123 (2014) Emerging understanding of multiscale tumor heterogene- 67. Vogelstein B, Kinzler KW (2015) The path to cancer –three strikes ity. Front Oncol 4:366 and you’re out. N Engl J Med 373(20):1895–1898 85. Ramon YCS, De Mattos-Arruda L, Sonenberg N, Cortes J, Peg V 68. Sehgal R, Sheahan K, O’Connell PR, Hanly AM, Martin ST, (2014) The intra-tumor heterogeneity of cell signaling factors in Winter DC (2014) Lynch syndrome: an updated review. Genes breast cancer: p4E-BP1 and peIF4E are diffusely expressed and (Basel) 5(3):497–507 are real potential targets. Clin Transl Oncol 16(11):937–941 69. Blanes A, Diaz-Cano SJ (2006) Complementary analysis of mi- 86. Gahete MD, Cordoba-Chacon J, Hergueta-Redondo M, Martinez- crosatellite tumor profile and mismatch repair defects in colorectal Fuentes AJ, Kineman RD, Moreno-Bueno G, Luque RM, Castano carcinomas. World J Gastroenterol 12(37):5932–5940 JP (2011) A novel human ghrelin variant (In1-ghrelin) and 70. Diaz-Cano SJ (2008) General morphological and biological fea- ghrelin-O-acyltransferase are overexpressed in breast cancer: po- tures of neoplasms: integration of molecular findings. tential pathophysiological relevance. PLoS One 6(8):e23302 Histopathology 53(1):1–19 87. Inoue K, Fry EA (2015) Aberrant splicing of estrogen receptor, 71. Yang Q, Yang Y, Zhou N, Tang K, Lau WB, Lau B, Wang W, Xu HER2, and CD44 genes in breast cancer. Genet Epigenet 7:19–32 L, Yang Z, Huang S et al (2018) Epigenetics in ovarian cancer: premise, properties, and perspectives. Mol Cancer 17(1):109 88. Castellana B, Aasen T, Moreno-Bueno G, Dunn SE, Ramon y 72. Shen H, Laird PW (2013) Interplay between the cancer genome Cajal S (2015) Interplay between YB-1 and IL-6 promotes the and epigenome. Cell 153(1):38–55 metastatic phenotype in breast cancer cells. Oncotarget 6(35): 73. Rastetter M, Schagdarsurengin U, Lahtz C, Fiedler E, Marsch W, 38239–38256 Dammann R, Helmbold P (2007) Frequent intra-tumoural hetero- 89. Silvera D, Formenti SC, Schneider RJ (2010) Translational control geneity of promoter hypermethylation in malignant melanoma. in cancer. Nat Rev Cancer 10(4):254–266 Histol Histopathol 22(9):1005–1015 90. Ramon y Cajal S, Missero C, Marchetti E, Dotto GP (1994) 74. Korshunova Y, Maloney RK, Lakey N, Citek RW, Bacher B, Dermal fibroblasts tumor suppression of ras-transformed Budiman A, Ordway JM, McCombie WR, Leon J, Jeddeloh JA keratinocytes is associated with induction of squamous cell differ- et al (2008) Massively parallel bisulphite pyrosequencing reveals entiation. Am J Pathol 145(4):846–855 the molecular complexity of breast cancer-associated cytosine- 91. Dotto GP, Weinberg RA, Ariza A (1988) Malignant transforma- methylation patterns obtained from tissue and serum DNA. tion of mouse primary keratinocytes by Harvey sarcoma virus and Genome Res 18(1):19–29 its modulation by surrounding normal cells. Proc Natl Acad Sci U 75. Varley KE, Mutch DG, Edmonston TB, Goodfellow PJ, Mitra RD S A 85(17):6389–6393 (2009) Intra-tumor heterogeneity of MLH1 promoter methylation 92. Perez-Villamil B, Romera-Lopez A, Hernandez-Prieto S, Lopez- revealed by deep single molecule bisulfite sequencing. Nucleic Campos G, Calles A, Lopez-Asenjo JA, Sanz-Ortega J, Acids Res 37(14):4603–4612 Fernandez-Perez C, Sastre J, Alfonso R et al (2012) Colon cancer 76. Moelans CB, Verschuur-Maes AH, van Diest PJ (2011) Frequent molecular subtypes identified by expression profiling and associ- promoter hypermethylation of BRCA2, CDH13, MSH6, PAX5, ated to stroma, mucinous type and different clinical behavior. PAX6 and WT1 in ductal carcinoma in situ and invasive breast BMC Cancer 12:260 cancer. J Pathol 225(2):222–231 93. Calon A, Lonardo E, Berenguer-Llergo A, Espinet E, Hernando- 77. De Carvalho DD, Sharma S, You JS, Su SF, Taberlay PC, Kelly Momblona X, Iglesias M, Sevillano M, Palomo-Ponce S, TK, Yang X, Liang G, Jones PA (2012) DNA methylation screen- Tauriello DV, Byrom D et al (2015) Stromal gene expression ing identifies driver epigenetic events of cancer cell survival. defines poor-prognosis subtypes in colorectal cancer. Nat Genet Cancer Cell 21(5):655–667 47(4):320–329 78. Brocks D, Assenov Y, Minner S, Bogatyrova O, Simon R, Koop 94. Zhang S, Jing Y, Zhang M, Zhang Z, Ma P, Peng H, Shi K, Gao C, Oakes C, Zucknick M, Lipka DB, Weischenfeldt J et al (2014) WQ, Zhuang G (2015) Stroma-associated master regulators of Intratumor DNA methylation heterogeneity reflects clonal evolu- molecular subtypes predict patient prognosis in ovarian cancer. tion in aggressive prostate cancer. Cell Rep 8(3):798–806 Sci Rep 5:16066 174 J Mol Med (2020) 98:161–177 95. Wang WQ, Liu L, Xu HX, Luo GP, Chen T, Wu CT, Xu YF, Xu J, 114. Krook MA, Chen HZ, Bonneville R, Allenby P, Roychowdhury S Liu C, Zhang B et al (2013) Intratumoral alpha-SMA enhances the (2019) Rapid research autopsy: piecing the puzzle of tumor het- prognostic potency of CD34 associated with maintenance of erogeneity. Trends Cancer 5(1):1–5 microvessel integrity in hepatocellular carcinoma and pancreatic 115. Bedard PL, Hansen AR, Ratain MJ, Siu LL (2013) Tumour het- cancer. PLoS One 8(8):e71189 erogeneity in the clinic. Nature 501(7467):355–364 96. Dotto GP (2014) Multifocal epithelial tumors and field 116. Siegel MB, He X, Hoadley KA, Hoyle A, Pearce JB, Garrett AL, cancerization: stroma as a primary determinant. J Clin Invest Kumar S, Moylan VJ, Brady CM, Van Swearingen AE et al (2018) 124(4):1446–1453 Integrated RNA and DNA sequencing reveals early drivers of 97. Meacham CE, Morrison SJ (2013) Tumour heterogeneity and metastatic breast cancer. J Clin Invest 128(4):1371–1383 cancer cell plasticity. Nature 501(7467):328–337 117. Hoadley KA, Siegel MB, Kanchi KL, Miller CA, Ding L, Zhao W, 98. Bissell MJ, Labarge MA (2005) Context, tissue plasticity, and He X, Parker JS, Wendl MC, Fulton RS et al (2016) Tumor evo- cancer: are tumor stem cells also regulated by the microenviron- lution in two patients with basal-like breast cancer: a retrospective ment? Cancer Cell 7(1):17–23 genomics study of multiple metastases. PLoS Med 13(12): 99. Polyak K (2011) Heterogeneity in breast cancer. J Clin Invest e1002174 121(10):3786–3788 118. Hoadley KA, Siegel MB, Kanchi KL, Miller CA, Ding L, Zhao W, 100. Polyak K, Haviv I, Campbell IG (2009) Co-evolution of tumor He X, Parker JS, Wendl MC, Fulton RS et al (2017) Correction: cells and their microenvironment. Trends Genet 25(1):30–38 tumor evolution in two patients with basal-like breast cancer: a 101. Ramon YCS, Capdevila C, Hernandez-Losa J, De Mattos-Arruda retrospective genomics study of multiple metastases. PLoS Med L, Ghosh A, Lorent J, Larsson O, Aasen T, Postovit LM, 14(1):e1002222 Topisirovic I (2017) Cancer as an ecomolecular disease and a 119. Mattos-Arruda LD, Sammut S-J, Ross EM, Bashford-Rogers R, neoplastic consortium. Biochim Biophys Acta 1868(2):484–499 Greenstein E, Morganella S, Rueda OM, Martinez-Saez E, Peg V, 102. Chapman A, Fernandez del Ama L, Ferguson J, Kamarashev J, Cortés J et al (2018) The integrated genomic and immune land- Wellbrock C, Hurlstone A (2014) Heterogeneous tumor subpopu- scapes of lethal metastatic breast cancer (MBC). J Clin Oncol lations cooperate to drive invasion. Cell Rep 8(3):688–695 36(15_suppl):1009–1009 120. Tabernero J, Lenz HJ, Siena S, Sobrero A, Falcone A, Ychou M, 103. Cheung KJ, Padmanaban V, Silvestri V, Schipper K, Cohen JD, Humblet Y, Bouche O, Mineur L, Barone C et al (2015) Analysis Fairchild AN, Gorin MA, Verdone JE, Pienta KJ, Bader JS et al of circulating DNA and protein biomarkers to predict the clinical (2016) Polyclonal breast cancer metastases arise from collective activity of regorafenib and assess prognosis in patients with met- dissemination of keratin 14-expressing tumor cell clusters. Proc astatic colorectal cancer: a retrospective, exploratory analysis of Natl Acad Sci U S A 113(7):E854–E863 the CORRECT trial. Lancet Oncol 16(8):937–948 104. Aceto N, Bardia A, Miyamoto DT, Donaldson MC, Wittner BS, 121. Malladi S, Macalinao DG, Jin X, He L, Basnet H, Zou Y, de Spencer JA, Yu M, Pely A, Engstrom A, Zhu H et al (2014) Stanchina E, Massague J (2016) Metastatic latency and immune Circulating tumor cell clusters are oligoclonal precursors of breast evasion through autocrine inhibition of WNT. Cell 165(1):45–60 cancer metastasis. Cell 158(5):1110–1122 122. Paul JM, Templeton SD, Baharani A, Freywald A, Vizeacoumar 105. Cheung KJ, Gabrielson E, Werb Z, Ewald AJ (2013) Collective FJ (2014) Building high-resolution synthetic lethal networks: a invasion in breast cancer requires a conserved basal epithelial ‘Google map’ of the cancer cell. Trends Mol Med 20(12):704–715 program. Cell 155(7):1639–1651 123. Blanes A, Diaz-Cano SJ (2006) DNA and kinetic heterogeneity 106. Martin-Pardillos A, Valls Chiva A, Bande Vargas G, Hurtado during the clonal evolution of adrenocortical proliferative lesions. Blanco P, Pineiro Cid R, Guijarro PJ, Hummer S, Bejar Serrano Hum Pathol 37(10):1295–1303 E, Rodriguez-Casanova A, Diaz-Lagares A et al (2019) The role 124. Blanes A, Rubio J, Martinez A, Wolfe HJ, Diaz-Cano SJ (2002) of clonal communication and heterogeneity in breast cancer. BMC Kinetic profiles by topographic compartments in muscle-invasive Cancer 19(1):666 transitional cell carcinomas of the bladder: role of TP53 and NF1 107. Keith B, Johnson RS, Simon MC (2011) HIF1alpha and genes. Am J Clin Pathol 118(1):93–100 HIF2alpha: sibling rivalry in hypoxic tumour growth and progres- 125. Blanes A, Rubio J, Sanchez-Carrillo JJ, Diaz-Cano SJ (2009) sion. Nat Rev Cancer 12(1):9–22 Coexistent intraurothelial carcinoma and muscle-invasive 108. Turajlic S, Swanton C (2016) Metastasis as an evolutionary pro- urothelial carcinoma of the bladder: clonality and somatic down- cess. Science 352(6282):169–175 regulation of DNA mismatch repair. Hum Pathol 40(7):988–997 109. Notta F, Mullighan CG, Wang JC, Poeppl A, Doulatov S, Phillips 126. Blanes A, Sanchez-Carrillo JJ, Diaz-Cano SJ (2006) Topographic LA, Ma J, Minden MD, Downing JR, Dick JE (2011) Evolution of molecular profile of pheochromocytomas: role of somatic down- human BCR-ABL1 lymphoblastic leukaemia-initiating cells. regulation of mismatch repair. J Clin Endocrinol Metab 91(3): Nature 469(7330):362–367 1150–1158 110. Ding L, Ley TJ, Larson DE, Miller CA, Koboldt DC, Welch JS, 127. Diaz-Cano SJ (2007) Kinetic topographical heterogeneity in fol- Ritchey JK, Young MA, Lamprecht T, McLellan MD et al (2012) licular thyroid neoplasms and growth patterns. Histopathology Clonal evolution in relapsed acute myeloid leukaemia revealed by 51(3):416–418 whole-genome sequencing. Nature 481(7382):506–510 128. Diaz-Cano SJ, Blanes A, Rubio J, Matilla A, Wolfe HJ (2000) 111. Janiszewska M, Liu L, Almendro V, Kuang Y, Paweletz C, Sakr Molecular evolution and intratumor heterogeneity by topographic RA, Weigelt B, Hanker AB, Chandarlapaty S, King TA et al compartments in muscle-invasive transitional cell carcinoma of (2015) In situ single-cell analysis identifies heterogeneity for the urinary bladder. Lab Investig 80(3):279–289 PIK3CA mutation and HER2 amplification in HER2-positive 129. Pozo L, Sanchez-Carrillo JJ, Martinez A, Blanes A, Diaz-Cano SJ breast cancer. Nat Genet 47(10):1212–1219 (2007) Differential kinetic features by tumour topography in cuta- 112. Bouwman P, Jonkers J (2014) Molecular pathways: how can neous small-cell neuroendocrine (Merkel cell) carcinomas. J Eur BRCA-mutated tumors become resistant to PARP inhibitors? Acad Dermatol Venereol 21(9):1220–1228 Clin Cancer Res 20(3):540–547 130. Diaz-Cano SJ (2000) Designing a molecular analysis of clonality 113. Marusyk A, Tabassum DP, Altrock PM, Almendro V, Michor F, in tumours. J Pathol 191(4):343–344 Polyak K (2014) Non-cell-autonomous driving of tumour growth 131. Diaz-Cano SJ (2001) Are PCR artifacts in microdissected samples supports sub-clonal heterogeneity. Nature 514(7520):54–58 preventable? Hum Pathol 32(12):1415–1416 J Mol Med (2020) 98:161–177 175 132. Diaz-Cano SJ, Blanes A, Wolfe HJ (2001) PCR techniques for 152. Jimenez-Sanchez A, Memon D, Pourpe S, Veeraraghavan H, Li Y, clonality assays. Diagn Mol Pathol 10(1):24–33 Vargas HA, Gill MB, Park KJ, Zivanovic O, Konner J et al (2017) 133. Nakamura T, Kuwai T, Kitadai Y, Sasaki T, Fan D, Coombes KR, Heterogeneous tumor-immune microenvironments among differ- Kim SJ, Fidler IJ (2007) Zonal heterogeneity for gene expression entially growing metastases in an ovarian cancer patient. Cell in human pancreatic carcinoma. Cancer Res 67(16):7597–7604 170(5):927–938 e920 134. Gonzalez-Garcia I, Sole RV, Costa J (2002) Metapopulation dy- 153. Graff JR, Konicek BW, Carter JH, Marcusson EG (2008) namics and spatial heterogeneity in cancer. Proc Natl Acad Sci U S Targeting the eukaryotic translation initiation factor 4E for cancer A 99(20):13085–13089 therapy. Cancer Res 68(3):631–634 135. Romo-Bucheli D, Janowczyk A, Gilmore H, Romero E, 154. Graff JR, Konicek BW, Lynch RL, Dumstorf CA, Dowless MS, Madabhushi A (2017) A deep learning based strategy for identi- McNulty AM, Parsons SH, Brail LH, Colligan BM, Koop JW et al fying and associating mitotic activity with gene expression de- (2009) eIF4E activation is commonly elevated in advanced human rived risk categories in estrogen receptor positive breast cancers. prostate cancers and significantly related to reduced patient sur- Cytometry A 91(6):566–573 vival. Cancer Res 69(9):3866–3873 136. Heng YJ, Lester SC, Tse GM, Factor RE, Allison KH, Collins LC, 155. Fan S, Ramalingam SS, Kauh J, Xu Z, Khuri FR, Sun SY (2009) Chen YY, Jensen KC, Johnson NB, Jeong JC et al (2017) The Phosphorylated eukaryotic translation initiation factor 4 (eIF4E) is molecular basis of breast cancer pathological phenotypes. J elevated in human cancer tissues. Cancer Biol Ther 8(15):1463– Pathol 241(3):375–391 1469 137. Trinh A, Trumpi K, De Sousa EMF, Wang X, de Jong JH, Fessler 156. Yoshizawa A, Fukuoka J, Shimizu S, Shilo K, Franks TJ, Hewitt E, Kuppen PJ, Reimers MS, Swets M, Koopman M et al (2017) SM, Fujii T, Cordon-Cardo C, Jen J, Travis WD (2010) Practical and robust identification of molecular subtypes in colo- Overexpression of phospho-eIF4E is associated with survival rectal cancer by immunohistochemistry. Clin Cancer Res 23(2): through AKT pathway in non-small cell lung cancer. Clin 387–398 Cancer Res 16(1):240–248 138. Louis DN, Perry A, Burger P, Ellison DW, Reifenberger G, von 157. Campbell L, Jasani B, Griffiths DF, Gumbleton M (2015) Deimling A, Aldape K, Brat D, Collins VP, Eberhart C et al (2014) Phospho-4e-BP1 and eIF4E overexpression synergistically drives International society of neuropathology–Haarlem consensus disease progression in clinically confined clear cell renal cell car- guidelines for nervous system tumor classification and grading. cinoma. Am J Cancer Res 5(9):2838–2848 Brain Pathol 24(5):429–435 158. Siddiqui N, Sonenberg N (2015) Signalling to eIF4E in cancer. 139. Yamamoto S, Maki DD, Korn RL, Kuo MD (2012) Radiogenomic Biochem Soc Trans 43(5):763–772 analysis of breast cancer using MRI: a preliminary study to define 159. Pettersson F, Del Rincon SV, Emond A, Huor B, Ngan E, Ng J, the landscape. AJR Am J Roentgenol 199(3):654–663 Dobocan MC, Siegel PM, Miller WH Jr (2015) Genetic and phar- 140. Jamshidi N, Jonasch E, Zapala M, Korn RL, Brooks JD, macologic inhibition of eIF4E reduces breast cancer cell migra- Ljungberg B, Kuo MD (2016) The radiogenomic risk score strat- tion, invasion, and metastasis. Cancer Res 75(6):1102–1112 ifies outcomes in a renal cell cancer phase 2 clinical trial. Eur 160. Musa J, Orth MF, Dallmayer M, Baldauf M, Pardo C, Rotblat B, Radiol 26(8):2798–2807 Kirchner T, Leprivier G, Grunewald TG (2016) Eukaryotic initia- 141. Eberwine J, Sul JY, Bartfai T, Kim J (2014) The promise of single- tion factor 4E-binding protein 1 (4E-BP1): a master regulator of cell sequencing. Nat Methods 11(1):25–27 mRNA translation involved in tumorigenesis. Oncogene 35(36): 142. Chappell L, Russell AJC, Voet T (2018) Single-cell (multi)omics 4675–4688 technologies. Annu Rev Genomics Hum Genet 19:15–41 161. Topisirovic I, Sonenberg N (2015) Translation and cancer. 143. Dal Molin A, Di Camillo B (2018) How to design a single-cell Biochim Biophys Acta 1849(7):751–752 RNA-sequencing experiment: pitfalls, challenges and perspec- 162. Rojo F, Najera L, Lirola J, Jimenez J, Guzman M, Sabadell MD, tives. Brief Bioinform. https://doi.org/10.1093/bib/bby007 Baselga J, Ramon y Cajal S (2007) 4E-binding protein 1, a cell 144. Zeng T, Dai H (2019) Single-cell RNA sequencing-based computa- signaling hallmark in breast cancer that correlates with pathologic tional analysis to describe disease heterogeneity. Front Genet 10:629 grade and prognosis. Clin Cancer Res 13(1):81–89 145. Cao J, Packer JS, Ramani V, Cusanovich DA, Huynh C, Daza R, 163. Tabassum DP, Polyak K (2015) Tumorigenesis: it takes a village. Qiu X, Lee C, Furlan SN, Steemers FJ et al (2017) Comprehensive Nat Rev Cancer 15(8):473–483 single-cell transcriptional profiling of a multicellular organism. 164. Solimini NL, Luo J, Elledge SJ (2007) Non-oncogene addiction Science 357(6352):661–667 and the stress phenotype of cancer cells. Cell 130(6):986–988 146. Dai W, Zhou F, Tang D, Lin L, Zou C, Tan W, Dai Y (2019) 165. Best MG, Sol N, Zijl S, Reijneveld JC, Wesseling P, Wurdinger T Single-cell transcriptional profiling reveals the heterogenicity in (2015) Liquid biopsies in patients with diffuse glioma. Acta colorectal cancer. Medicine (Baltimore) 98(34):e16916 Neuropathol 129(6):849–865 147. Luo J, Solimini NL, Elledge SJ (2009) Principles of cancer thera- 166. De Mattos-Arruda L, Caldas C (2015) Cell-free circulating tumour py: oncogene and non-oncogene addiction. Cell 136(5):823–837 DNA as a liquid biopsy in breast cancer. Mol Oncol. https://doi. 148. Khirade MF, Lal G, Bapat SA (2015) Derivation of a fifteen gene org/10.1016/j.molonc.2015.12.001 prognostic panel for six cancers. Sci Rep 5:13248 167. Alix-Panabieres C, Pantel K (2016) Clinical applications of circu- 149. Werner HM, Mills GB, Ram PT (2014) Cancer systems biology: a lating tumor cells and circulating tumor DNA as liquid biopsy. peek into the future of patient care? Nat Rev Clin Oncol 11(3): Cancer Discov. https://doi.org/10.1158/2159-8290.CD-15-1483 167–176 168. De Mattos-Arruda L, Cortes J, Santarpia L, Vivancos A, 150. Gerlinger M, Quezada SA, Peggs KS, Furness AJ, Fisher R, Tabernero J, Reis-Filho JS, Seoane J (2013) Circulating tumour Marafioti T, Shende VH, McGranahan N, Rowan AJ, Hazell S cells and cell-free DNA as tools for managing breast cancer. Nat et al (2013) Ultra-deep T cell receptor sequencing reveals the Rev Clin Oncol 10(7):377–389 complexity and intratumour heterogeneity of T cell clones in renal 169. Leary RJ, Kinde I, Diehl F, Schmidt K, Clouser C, Duncan C, cell carcinomas. J Pathol 231(4):424–432 Antipova A, Lee C, McKernan K, De La Vega FM et al (2010) 151. Reuben A, Spencer CN, Prieto PA, Gopalakrishnan V, Reddy SM, Development of personalized tumor biomarkers using massively Miller JP, Mao X, De Macedo MP, Chen J, Song X et al (2017) parallel sequencing. Sci Transl Med 2(20):20ra14 Genomic and immune heterogeneity are associated with differen- 170. Leary RJ, Sausen M, Kinde I, Papadopoulos N, Carpten JD, Craig tial responses to therapy in melanoma. NPJ Genom Med 2 D, O'Shaughnessy J, Kinzler KW, Parmigiani G, Vogelstein B et al 176 J Mol Med (2020) 98:161–177 (2012) Detection of chromosomal alterations in the circulation of 185. Shao H, Chung J, Balaj L, Charest A, Bigner DD, Carter BS, cancer patients with whole-genome sequencing. Sci Transl Med Hochberg FH, Breakefield XO, Weissleder R, Lee H (2012) 4(162):162ra154 Protein typing of circulating microvesicles allows real-time mon- 171. Forshew T, Murtaza M, Parkinson C, Gale D, Tsui DW, Kaper F, itoring of glioblastoma therapy. Nat Med 18(12):1835–1840 Dawson SJ, Piskorz AM, Jimenez-Linan M, Bentley D et al 186. Chen WW, Balaj L, Liau LM, Samuels ML, Kotsopoulos SK, (2012) Noninvasive identification and monitoring of cancer mu- Maguire CA, Loguidice L, Soto H, Garrett M, Zhu LD et al tations by targeted deep sequencing of plasma DNA. Sci Transl (2013) BEAMing and droplet digital PCR analysis of mutant Med 4(136):136ra168 IDH1 mRNA in glioma patient serum and cerebrospinal fluid 172. Murtaza M, Dawson SJ, Tsui DW, Gale D, Forshew T, Piskorz extracellular vesicles. Mol Ther Nucleic Acids 2:e109 AM, Parkinson C, Chin SF, Kingsbury Z, Wong AS et al (2013) 187. Best MG, Sol N, Kooi I, Tannous J, Westerman BA, Rustenburg F, Non-invasive analysis of acquired resistance to cancer therapy by Schellen P, Verschueren H, Post E, Koster J et al (2015) RNA-Seq sequencing of plasma DNA. Nature 497(7447):108–112 of tumor-educated platelets enables blood-based pan-cancer, 173. Dawson SJ, Tsui DW, Murtaza M, Biggs H, Rueda OM, Chin SF, multiclass, and molecular pathway cancer diagnostics. Cancer Dunning MJ, Gale D, Forshew T, Mahler-Araujo B et al (2013) Cell 28(5):666–676 Analysis of circulating tumor DNA to monitor metastatic breast 188. Pan W, Gu W, Nagpal S, Gephart MH, Quake SR (2015) Brain cancer. N Engl J Med 368(13):1199–1209 tumor mutations detected in cerebral spinal fluid. Clin Chem 174. De Mattos-Arruda L, Weigelt B, Cortes J, Won HH, Ng CK, 61(3):514–522 Nuciforo P, Bidard FC, Aura C, Saura C, Peg V et al (2014) 189. Wang Y, Springer S, Zhang M, McMahon KW, Kinde I, Dobbyn Capturing intra-tumor genetic heterogeneity by de novo mutation L, Ptak J, Brem H, Chaichana K, Gallia GL et al (2015) Detection profiling of circulating cell-free tumor DNA: a proof-of-principle. of tumor-derived DNA in cerebrospinal fluid of patients with pri- Ann Oncol 25(9):1729–1735 mary tumors of the brain and spinal cord. Proc Natl Acad Sci U S 175. Rothe F, Laes JF, Lambrechts D, Smeets D, Vincent D, Maetens A 112(31):9704–9709 M, Fumagalli D, Michiels S, Drisis S, Moerman C et al (2014) 190. Shi R, Wang PY, Li XY, Chen JX, Li Y, Zhang XZ, Zhang CG, Jiang Plasma circulating tumor DNA as an alternative to metastatic bi- T, Li WB, Ding W et al (2015) Exosomal levels of miRNA-21 from opsies for mutational analysis in breast cancer. Ann Oncol 25(10): cerebrospinal fluids associated with poor prognosis and tumor recur- 1959–1965 rence of glioma patients. Oncotarget 6(29):26971–26981 176. Bettegowda C, Sausen M, Leary RJ, Kinde I, Wang Y, Agrawal N, 191. Touat M, Duran-Pena A, Alentorn A, Lacroix L, Massard C, Bartlett BR, Wang H, Luber B, Alani RM et al (2014) Detection of Idbaih A (2015) Emerging circulating biomarkers in glioblastoma: circulating tumor DNA in early- and late-stage human malignan- promises and challenges. Expert Rev Mol Diagn 15(10):1311– cies. Sci Transl Med 6(224):224ra224 1323 177. Olsson E, Winter C, George A, Chen Y, Howlin J, Tang MH, 192. Bardelli A, Pantel K (2017) Liquid biopsies, what we do not know Dahlgren M, Schulz R, Grabau D, van Westen D et al (2015) (yet). Cancer Cell 31(2):172–179 Serial monitoring of circulating tumor DNA in patients with pri- 193. Diaz LA Jr, Williams RT, Wu J, Kinde I, Hecht JR, Berlin J, Allen mary breast cancer for detection of occult metastatic disease. B, Bozic I, Reiter JG, Nowak MA et al (2012) The molecular EMBO Mol Med. https://doi.org/10.15252/emmm.201404913 evolution of acquired resistance to targeted EGFR blockade in 178. De Mattos-Arruda L, Mayor R, Ng CK, Weigelt B, Martinez- colorectal cancers. Nature 486(7404):537–540 Ricarte F, Torrejon D, Oliveira M, Arias A, Raventos C, Tang J 194. De Mattos-Arruda L, Caldas C (2016) Cell-free circulating tumour et al (2015) Cerebrospinal fluid-derived circulating tumour DNA DNA as a liquid biopsy in breast cancer. Mol Oncol 10(3):464–474 better represents the genomic alterations of brain tumours than plasma. Nat Commun 6:8839 195. Siravegna G, Marsoni S, Siena S, Bardelli A (2017) Integrating 179. Murtaza M, Dawson SJ, Pogrebniak K, Rueda OM, Provenzano liquid biopsies into the management of cancer. Nat Rev Clin E, Grant J, Chin SF, Tsui DW, Marass F, Gale D et al (2015) Oncol 14(9):531–548 Multifocal clonal evolution characterized using circulating tumour 196. Abbosh C, Birkbak NJ, Wilson GA, Jamal-Hanjani M, Constantin DNA in a case of metastatic breast cancer. Nat Commun 6:8760 T, Salari R, Le Quesne J, Moore DA, Veeriah S, Rosenthal R et al 180. Newman AM, Bratman SV, To J, Wynne JF, Eclov NC, Modlin (2017) Phylogenetic ctDNA analysis depicts early-stage lung can- LA, Liu CL, Neal JW, Wakelee HA, Merritt RE et al (2014) An cer evolution. Nature 545(7655):446–451 ultrasensitive method for quantitating circulating tumor DNA with 197. Litjens G, Sanchez CI, Timofeeva N, Hermsen M, Nagtegaal I, broad patient coverage. Nat Med 20(5):548–554 Kovacs I, Hulsbergen-van de Kaa C, Bult P, van Ginneken B, van 181. Siravegna G, Mussolin B, Buscarino M, Corti G, Cassingena A, der Laak J (2016) Deep learning as a tool for increased accuracy Crisafulli G, Ponzetti A, Cremolini C, Amatu A, Lauricella C et al and efficiency of histopathological diagnosis. Sci Rep 6:26286 (2015) Clonal evolution and resistance to EGFR blockade in the 198. Mobadersany P, Yousefi S, Amgad M, Gutman DA, Barnholtz- blood of colorectal cancer patients. Nat Med 21(7):827 Sloan JS, Velazquez Vega JE, Brat DJ, Cooper LAD (2018) 182. Newman AM, Lovejoy AF, Klass DM, Kurtz DM, Chabon JJ, Predicting cancer outcomes from histology and genomics using Scherer F, Stehr H, Liu CL, Bratman SV, Say C et al (2016) convolutional networks. Proc Natl Acad Sci U S A 115(13): Integrated digital error suppression for improved detection of circu- E2970–E2979 lating tumor DNA. Nat Biotechnol. https://doi.org/10.1038/nbt.3520 199. Bychkov D, Linder N, Turkki R, Nordling S, Kovanen PE, Verrill 183. Girotti MR, Gremel G, Lee R, Galvani E, Rothwell D, Viros A, C, Walliander M, Lundin M, Haglund C, Lundin J (2018) Deep Mandal AK, Lim KH, Saturno G, Furney SJ et al (2016) learning based tissue analysis predicts outcome in colorectal can- Application of sequencing, liquid biopsies, and patient-derived cer. Sci Rep 8(1):3395 xenografts for personalized medicine in melanoma. Cancer 200. Michaut M, Chin SF, Majewski I, Severson TM, Bismeijer T, de Discov 6(3):286–299 Koning L, Peeters JK, Schouten PC, Rueda OM, Bosma AJ et al 184. Lohr JG, Adalsteinsson VA, Cibulskis K, Choudhury AD, (2016) Integration of genomic, transcriptomic and proteomic data Rosenberg M, Cruz-Gordillo P, Francis JM, Zhang CZ, Shalek identifies two biologically distinct subtypes of invasive lobular AK, Satija R et al (2014) Whole-exome sequencing of circulating breast cancer. Sci Rep 6:18517 tumor cells provides a window into metastatic prostate cancer. Nat 201. Gilmore SJ (2018) Automated decision support in melanocytic Biotechnol 32(5):479–484 lesion management. PLoS One 13(9):e0203459 J Mol Med (2020) 98:161–177 177 202. Yu KH, Zhang C, Berry GJ, Altman RB, Re C, Rubin DL, Snyder M 204. Horning SJ (2017) A new cancer ecosystem. Science 355(6330): (2016) Predicting non-small cell lung cancer prognosis by fully auto- 1103 mated microscopic pathology image features. Nat Commun 7:12474 203. Djuric U, Zadeh G, Aldape K, Diamandis P (2017) Precision his- Publisher’s note Springer Nature remains neutral with regard to jurisdic- tology: how deep learning is poised to revitalize histomorphology tional claims in published maps and institutional affiliations. for personalized cancer care. NPJ Precis Oncol 1(1):22