Cancer evolution and heterogeneity

Abstract Undoubtedly, intratumor heterogeneity (ITH) is one of the causes of the intractability of cancers. Recently, technological innovation in genomics has promoted studies on ITH in solid tumors and on the pattern and level of diversity, which varies among malignancies. We profiled the genome in multiple regions of nine colorectal cancer (CRC) cases. The most impressive finding was that in the late phase, a parental clone branched into numerous subclones. We found that minor mutations were dominant in advanced CRC named neutral evolution; that is, driver gene aberrations were observed with high proportion in the early‐acquired phase, but low in the late‐acquired phase. Then, we validated that neutral evolution could cause ITH in advanced CRC by super‐computational analysis. According to the clinical findings, we explored a branching evolutionary process model in cancer evolution, which assumes that each tumor cell has cellular automaton. According to the model, we verified factors to foster ITH with neutral evolution in advanced CRC. In this review, we introduce recent advances in the field of ITH including the general component of ITH, clonal selective factors that consolidate the evolutionary process, and a representative clinical application of ITH.


| INTRODUCTION
In general, intratumor heterogeneity (ITH) is considered one of the critical causes of intractability in the treatment of cancers; therefore, it is very important to clarify the precise mechanism underlying ITH to establish a strategy for the treatment of solid cancers.
Recently, ITH-related studies have used next-generation sequencing to conduct whole-exome sequencing of multiple excised samples from primary and/or metastatic tumors and have comprehensively integrated whole sequence data (Table 1).  This multiregional analysis (MRA) sequencing approach enabled us not only to observe spatial heterogeneity, but also to calculate temporal alterations and eventually disclose the evolution of tumors. There are two types of somatic aberration in a tumor: ubiquitous aberrations (founder mutations, trunk mutations, or clonal mutations) and scattered aberrations (progressor mutations, branch/leaf mutations, or subclonal mutations). The former and the latter are triggered by a carcinogenic event and a late event, respectively.
In addition, we disclose the clinical significance of defining the clonality of genomic aberrations by the MRA method from the viewpoint of targeting the cancers. In the phylogenic tree of ITH in the study of cancer evolution, clonal mutations were located in the trunk, and minor mutations were in branches and leaves. According sufficient antitumor effect, we should target any clonal events in the trunk for eliminating cancer.
In this review, we update the general information of ITH as follows: (i) components of ITH; (ii) evolution model for chronological factors for ITH; (iii) clonal selective factors for fostering ITH; and (iv) a representative large study of the clinical applications of ITH.

| Mutation spectrum
The degree of temporal and spatial heterogeneity depends on the malignancy. In general, melanoma and lung cancer accumulated a large number of somatic nucleotide variants (SNV); however, the diversity level was low and SNV were relatively ubiquitous in the tumor. Somatic mutations are present in all cells and they are the consequence of multiple mutational processes, including the intrinsic slight infidelity of the DNA replication machinery, exogenous or endogenous mutagen exposures, and enzymatic modification of DNA and, at present, they have been classified into 30 signatures. [35][36][37][38] The nucleotide substitutions in those tumors were characterized as C>T from ultraviolet light (signature 7) 23 and C>A from smoking (signature 4). 19,20 Both substitutions were observed as founder events. These findings indicated that strong outer mutagens, such as UV light and cigarettes, may be carcinogens. In addition, lowgrade glioma showed an exacerbated diversity in ITH after treatment with the alkylating agent temozolomide. 39 In non-small-cell lung cancers (NSCLC) and bladder cancer, the apolipoprotein B mRNA editing enzyme, catalytic polypeptide-like (APOBEC) family of proteins was associated with fostering many subclones (signature 2). 19,20,40

| Causative mechanism of ITH
There is a dispute between natural (Darwinian) selection and neutral evolution. In renal cell carcinoma, driver genes, such as mTOR, TSC1, PTEN, and PIK3CA, were observed at subclonal or parallel positions in an identical tumor. 31,32 These alterations were considered to be a result of natural selection. On the contrary, progressor alterations accumulated in few driver genes, and the passenger mutations fostered neutral evolution.
According to the distribution of variant allele frequency (VAF), neutral evolution was observed in 30% of all malignancies. 41 Our previous wholeexome sequencing (WES) study by MRA showed neutral evolution in advanced CRC cases, which was validated by computational simulation analysis. According to our previous model, accumulation of non-driver genes, presence of cancer stem cells, and the microenvironment around cancer cells can foster neutral evolution in advanced CRC. 11

| EVOLUTION MODEL AND CHRONOLOGICAL FACTORS TO FORM ITH
Our multiregional sequencing study showed that progressor mutations comprised 40% of all mutations, and most of them were classified as passenger mutations and form ITH. Neutral evolution along T A B L E 1 Achievements in the field of intratumor heterogeneity and evolutions of solid cancers F I G U R E 1 Branching evolutionary process (BEP) model. A, A cell has n genes, d of which are driver genes. In a unit time step, a cell divides and dies with probabilities p and q, respectively. A cell division mutates each gene with a probability r. One driver mutation increases p by ffold. In this model, f indicates strength of the driver genes. B, Population entropy depends on parameters d and f. The division probability increases per driver mutation. Red area indicates negentropy or syntropy, whereas white area indicates entropy. C, Existence of strong driver genes leads to a homogenous tumor. D, Multiple driver genes of moderate strength generate intratumor heterogeneity with clonal evolution is a principal cause of ITH and fosters advanced CRC. 11 We simulated heterogeneous cancer evolution as "branching evolutionary process (BEP) model" by supercomupter ( Figure 1). In this model, each cell gradually accumulates driver mutations as well as accompanying passenger mutations, which do not affect the cell division rate and, finally, a tumor is formed with numerous accumulated mutations. According to the model, we found that mutations in driver genes were clonal, and non-driver genes were subclonal; therefore, advanced CRC showed ITH not by natural

| Chemotherapy and treatment
According to the MRA of bladder cancer and ovarian cancer, there were several cases in which anticancer drugs and hormonal agents might determine the clones that survive. Treatment of cancer might be one of the selective pressures on clones, and recurrence might be derived from surviving clones. 7,33 As depicted in Figure 2, we implemented the simulation study to prove the presence of ITH with selective mutations in driver genes by exposing four environmental pressures, such as chemotherapy and other treatment modalities.
Existence of environmental selection can also enhance intratumor heterogeneity which is shaping the real heterogenous tumor.  Clones carrying passenger mutations would senesce or die, such that the mutation would be lost from the catalog of variants seen in resected cancer specimens. This is negative or purifying selection, which leads to a dN/dS <1 in a given gene or set of genes, if it occurs at appreciable rates. [47][48][49][50] If the negatively selected genes are 0.02%-0.5% of all genes, the dN/dS <1, and clones with coding mutations will be lost per tumor. On the contrary, some somatic mutations, such as driver genes, can confer a growth advantage, whereas others may impair cell survival or proliferation. Positively selected genes have a dN/dS ≥1 and were 1%-3.9% of all genes. There were 1-10 + driver mutations per tumor, and they had a much stronger force than negative selection.

| Cytokines in the microenvironment
According to the study by Martincorena et al, 50 CRC showed a relatively higher dN/dS ratio than other cancers. Therefore, the dN/dS ratio indicated that positively selected genes might exist under negative selection in advanced CRC. However, our previous study disclosed neutral evolution in advanced CRC with a number of minor (non-driver) mutations according to our multisampling and sequence data. 11 To comprehend this contradiction, we have provided the following possible explanation.

| Impaired neoantigen presentation by chromosomal aberration
Immune evasion is one of the hallmarks of cancer. Losing the ability to present neoantigens, as a result of human leukocyte antigen

CLINICAL APPLICATIONS
To

DISCLOSURE
There is no relevant financial or nonfinancial relationships to disclose.
F I G U R E 2 A, Implementation of environmental selection (n, number of genes; d, number of driver genes). If mutation has occurred in each quadrant of the tumor, selective driver genes increase growth rate. B, Existence of environmental selection can also enhance intratumor heterogeneity, which looks close to the actual heterogenous tumor