How many mutations are required to produce a human cancer cell?

Assessment of theoretical models and their experimental support

Bachelor Thesis 2011 35 Pages

Biology - General, Basics




How many mutations are required to produce a human cancer cell?
Sequence data
In vitro data

Is genetic instability necessary to acquire sufficient mutations?
The mutator phenotype hypothesis
Arguments which undermine the calculations’ assumptions
Clonal evolution and natural selection
Arguments for the calculations validity
Tissue Biology
CpG island promoter hypermethylation
Global CpG hypomethylation

Does genetic instability accelerate tumour progression?
Cell clone ecology hypothesis
Mathematical assessment
Lab based test
Clinical data
Sequence Data
Implications for therapy



Ordinarily, innate biochemical circuits ensure that physiological control of cellular proliferation is maintained for the benefit of the entire organisms’ fitness. Tumour progression involves the stepwise rewiring of these circuits to alter these proliferative controls for the short term fitness benefit of the individual clone. This allows it to outcompete its neighbours in what is accepted to be a process analogous to Darwinian natural selection. There are numerous routes that can be taken to bring about these re wirings, but as Hanahan and Weinberg (2000) suggest, they are all different manifestations of the same six fundamental hallmarks of cancer (Fig .1.).

It has been suggested that if multiple genetic alterations are required to bring about the above mentioned changes, the low background level of genetic instability may be prohibitive to tumorigenesis. Loeb (1991) postulated that increased genetic instability could account for cancer incidence rates by accelerating the generation of variation. Other authors (Bodmer 2008) have stressed that these models ignore the power of natural selection, which according to his groups models, can account for the genetic changes without increased genetic instability.

However, the multistep Darwinian model alone is too simplistic. As is discussed below, advances in tissue biology suggest the actual population undergoing selection may be limited to a small sub population at the tip of a cellular hierarchy. If this is the case, it drastically reduces the variability available for selection to act on, potentially increasing the time between the “steps” of tumorigenesis. These observations have even been seen by some (Weinberg 2007) to argue for the addition of an additional hallmark: genetic instability to the six already suggested in Fig.1. as an accelerating mechanism. alterations may not only offer a complimentary mechanism for generating variability but potentially an alternative route for initiating and sustaining tumour progression.

illustration not visible in this excerpt

Fig. 1.The six hallmarks of cancer suggested to be necessary for a cell to progress to malignancy together with an example for each pathway:

1. Growth signal autonomy: normal cells rely on a limited supply of growth factors, one of the ways cancers can evade this is to gain ‘cellular autonomy’ e.g. point mutation of Ras protoncogene results in proliferation stimulated constitutively via the SOS Ras Raf MAPK pathway.
2. Insensitivity to anti growth signals: e.g. by knock out or missense mutation of receptors for antigrowth signals such as TGF b or knockout of downstream signalling molecules e.g. pRb
3. Invasion and Metastasis: this step is important as it accounts for around 90% of cancer deaths and generally involves mutations in or shifts in cell cell adherence molecules (CAM’s) e.g. E cadherin and changes in ECM degrading/remodelling protease expression/secretion.
4. Replicative immortality: 85 90% of tumors have increased telomerase expression
5. Angiogenesis: due to the requirement for nutrients no cell should be further than 100um from a capillary, for the cancer to grow any larger than this it must recruit capillaries. It does this by shifting the balance of the “angiogenic switch” to the The plot thickens further with the advent of the epigenetic era. Epigenetic proangiogenic side.
6. Evading apoptosis: e.g. mutation of the “guardian of the genome” p53. (Diagram reproduced from Hanahan and Weinberg. 2000)

Due to the many variables and heterogeneous nature of cancer, no step in gaining an understanding of it is likely to be straightforward. The question of whether genetic instability is necessary or not will undoubtedly be the same. Since much research has been conducted in this area a theoretical approach should provide some answers to the question of whether genetic instability is necessary.

The approach taken here is to fractionate the problem into its constituent parts and then attempt to find an answer to each individually. When combined, the answers should lead to an overall greater understanding of the question and its answer.

How many mutations are required to produce a human cancer cell?

Is genetic instability necessary to acquire sufficient mutations?

Does raised genetic instability accelerate tumour progression, and is this the mechanism by which human cancer actually evolves?

The theoretical approach can contribute to every step of scientific enquiry involved in answering the question (Fig.2.): in drawing conclusions from the research, in the generation of hypotheses, models, and potentially in generating predictions which can be tested, thus guiding further research. The conclusions reached by this theoretical synthesis should help guide future research into cancer therapies and prevention strategies.

illustration not visible in this excerpt

Fig. 2. Role of theoretical approaches in scientific inquiry: Theoretical methods can contribute substantially at every step although it depends on experimental verification. (Reproduced from Beckman and Loeb. 2005)

A critical parameter which must be considered when trying to assess whether genetic instability is necessary for human cancer is the number of genetic mutations which are necessary to acquire the six hallmarks of cancer described in the last section.

Sequence data

The abundance of data from large scale sequencing has revealed a great deal about the armoury from which cancers choose their arsenal of mutations in the somatic evolutionary arms race. A recent study of a range of cancer types by Greenman et al. (2007) revealed 1000 point mutations in the coding regions of 518 genes alone. By looking at the effects of differences in selection pressures, they were able to discern that a surprisingly high number, 120 of the 518 genes were “drivers“. However, even when the problem of discerning driver mutations has been overcome, these data only give the average number of mutations which provide a selective advantage. Many of these may be “tweaks“ of the genome of the original founding cancer cell which may give the cancer more of a growth advantage, but are not strictly necessary.

This approach is therefore too simplistic. The crucially important question for this discussion is: what is the minimum complement of genetic changes from the above arsenal which is sufficient to produce a cancer cell? Below is presented a summary of the data from several independent lines of investigation which are now converging to come closer to providing an answer to this question.


Cook et al (1969) and Armitage and Doll (1985) used cancer incidence statistics from

a wide range of sources to show that different cancers had a range of best fit curves with the majority of gradients ranging from 3.5 to 6.5. This made it clear there would be no single answer to the question of how many mutations are required to reach the tumorigenic phenotype and generalising across type (as had been done till then) might be a simplistic way of approaching the problem.

More recently W.D Stein et al. (1990), taking cycles of mutation, clonal expansion and selection into account, developed and rigorously tested a simplified version of a two stage model (originally put forward by Moolgavkar and Knudson (1981)) against a wide range of epidemiological data. They argue that in the great majority of cases their data are consistent with at least three mutational events being required to bring about the progression of cancer through the two hypothetical stages.

Thus, the above examples, together with a wider review of the literature (Moolgavkar and Knudson 1981; Lubeck and Moolgavkar 2002), point to between 2 and 7 rate limiting steps leading to a clinically apparent cancer. However this has also highlighted the weaknesses of mathematical analysis of epidemiological data. It is often found that a variety of models can fit the same data, and even if the correct model is found, each rate determining step does not necessarily correspond to the number of mutations.

In vitro data

In vitro transformation assays in the Weinberg lab using viral vectors and oncoproteins have been used to try and distil down the distinct set of changes corresponding to the above steps. Hahn et al. (1999) found that disrupting a minimum of five pathways this way: Ras, p53, pRb, PP2A and telomerase could generate tumorigenic cells. Rangarajan and Weinberg (2004) have shown requirements not only vary between murine and human cells but also between different human cells. They found human fibroblasts specifically needed: p53, pRb, PP2A, telomerase, Raf, and Ral GEFs whereas human embryonic kidney cells required activation of RalGEFs and PI3 kinase but not Raf. These results raise questions about the accuracy of murine models and the practice of attempting to generalise requirements for tumorigenesis across tissue type.

Thus, in Vitro experiments hesitantly point to a minimum of ~five pathways (Weinberg 2007) needing to be disrupted in the majority of cancers which could require five mutations. However, the assay used by Hahn et al. only tested for tumorigenic ability to form benign tumours and it is likely more pathways need to be disrupted to progress to full malignancy. The recessive nature of many of the disruptions in the pathways mentioned further increases the uncertainty as to how many mutagenic steps there are in tumorigenesis.


illustration not visible in this excerpt

Fig. 4. Major genetic alterations associated with colorectal tumorigenesis. (Adapted from Jones et al. 2008)

Pathological analyses have long been seen to point to tumour progression occurring as a multistep process (Foulds 1954). Colorectal tumorigenesis has long been known to proceed through a set of well defined clinical stages, each of which is thought to arise due to a characteristic mutation. In a step towards integrating the in vitro work with the actual aetiology of cancer, work by Fearon and Vogelstein (1990) resulted in a speculative model of the actual genetic changes occurring during tumour progression. This has been updated and somewhat expanded by Jones et al (2008) using more accurate genetic sequencing (Fig.4.). Such models give some credence to the in vitro data and numbers derived thereof above, but it is clear that this sequence is only a single possible route of many and the original data has been seen by some to suggest more steps.

This section has demonstrated that despite large amounts of research having been done in this area, we are still far from making a confident assessment of what the minimum number of genetic mutations is for reprogramming physiologically controlled somatic cells to pathologically deranged cancer cells. Epidemiology suggests between 2 and 7 rate limiting stochastic events. In vitro assessment, to some extent corroborated by pathological observation, suggests these could correspond to the disruption of approximately five biochemical pathways, although it is clear that this requirement varies both spatially and temporally in the body. Therefore a commonly used (Hanahan and Weinberg (2000); Rangarajan et al (2004); Beckman and Loeb 2005) number for roughly estimating the number of mutations needed for the majority of cancers is ~6, although as has been shown, this can clearly vary by a great deal.



ISBN (eBook)
ISBN (Book)
File size
1018 KB
Catalog Number
Institution / College
University of Cambridge
Cancer; genetic instability; Mutation;




Title: How many mutations are required to produce a human cancer cell?