Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 97 (3), 1236-41

Genetic Reconstruction of Individual Colorectal Tumor Histories

Affiliations

Genetic Reconstruction of Individual Colorectal Tumor Histories

J L Tsao et al. Proc Natl Acad Sci U S A.

Abstract

It is difficult to observe human tumor progression as precursor lesions are systematically removed. Alternatives to direct observations, commonly used to reveal the hidden past of species and populations, are sequence comparisons or molecular clocks. Noncoding microsatellite (MS) loci were employed as molecular tumor clocks in 13 human mutator phenotype (MSI(+)) colorectal tumors. Quantitative analysis revealed that specific patterns of somatic MS mutations accumulate with division after loss of mismatch repair (MMR). Tumors had unique patterns of MS mutation, and, therefore, based on this model, each tumor had its own unique history. Loss of MMR occurred very early relative to terminal clonal expansion, with an estimated average of 2,300 divisions since loss of MMR and 280 divisions since expansion. Contrary to the classical adenoma-cancer sequence, MSI(+) adenomas were nearly as old as cancers (2,000 versus 2,400 divisions since loss of MMR). Negative clinical examinations preceded six tumors, independently documenting an absence of visible precursors during early MSI(+) adenoma or cancer progression. These findings further extend a window beyond visible progression since loss of MMR appears to start a genetic phase involving clone sizes or phenotypes below a threshold of clinical detection. This previously occult prologue before visible neoplasia is longer and therefore likely more important than generally appreciated.

Figures

Figure 1
Figure 1
(a) A tumor history is “read” from the mutations present in its cells. All of these cells are related to a single founder cell that represents the final bottleneck along a progression pathway. The history of this founder cell can be traced back along a single lineage because of the bottleneck nature of progression. We define the start of this pathway as the somatic loss of MMR. MS loci can record this history. Consider a single MS locus. After loss of MMR, the locus will randomly become larger or smaller, with the difference from its germline length a function of the number of divisions since loss of MMR. Upon terminal clonal expansion, the locus will become polymorphic. The time since terminal clonal expansion is reflected in the width or variance (Salleles2) of the tumor allele frequency distribution. The MS allele in the founder cell can be inferred by the most common allele present in the tumor. The time since loss of MMR preceding the bottleneck is reflected in the difference (Δgermline) between the length of the germline allele and the founder allele. Although the stochastic nature of mutation makes a single MS locus relatively uninformative, the analysis repeated at 20–30 different loci is robust. (b) Simulations of MS mutation. Data represents the results of 1,000 trials with 20–30 MS loci, and a symmetric stepwise model with the chance of addition of one repeat of 0.0025, and loss of one repeat of 0.0025, with the total mutation rate of 0.005 per division. In each scenario, the final tumor size is 1.0 cm3 or one billion cells. Different patterns of MS mutations, summarized by the variances, Salleles2 and Sloci2, are obtained with identical numbers of divisions (2,000) but different tumor histories. Therefore, a history of a human tumor can be inferred by sampling its MS alleles and estimating Salleles2 and Sloci2. (c) Example of simulations with identical clonal expansion histories (20 terminal exponential divisions) but different times since loss of MMR. The average value of Sloci2 increases with the numbers of divisions since loss of MMR whereas Salleles2 is constant. The graph illustrates the mean and 95% confidence intervals (dotted lines) of the simulations. The triangles in the graph represent values obtained from 210- to 352-day-old tissue culture experiments that mirror these simulations (see Fig. 2).
Figure 2
Figure 2
A single HCT 116 cell was isolated, grown, and subsequently split into different sublines. After 190–332 days, single clones were isolated from each subclone, were expanded for 20 more divisions, and were typed at 24 MS loci (autoradiographs from three 210-day-old sublines and six loci are illustrated). Relative to germline (open circles or the allele size of the original clone), subline alleles are randomly the same, or larger or smaller (filled circles). The distribution of these changes (Δgermline) and their estimated variances (Sloci2) are small compared with human tumors (compare with Fig. 3). Variation between alleles at a single locus is minimal because only 20 divisions occurred during terminal expansions (Salleles2 ≈ 0). This experiment mimics five simulated trials with a history of 210 divisions since loss of MMR and a terminal clonal expansion of 20 exponential divisions (see Fig. 1c). The experimental Sloci2 values are consistent with the simulations (see triangles in the graph of Fig. 1c).
Figure 3
Figure 3
(a) Autoradiographs of human tumor MS alleles. After dilution and PCR, the germline alleles (open circles) and tumor specific alleles (filled circles) become evident. Although the tumor alleles are polymorphic, they exhibit a modal size [−2 for DXS997 in the invasive cancer of Patient V (left) and −5 for DXS8011 in the right adenoma region of Patient II (right)]. (b) The differences from germline (Δgermline) of the 21–30 different MS loci from Patients I or II. Broader distributions, summarized by their variances (Sloci2 in parentheses), indicate greater numbers of divisions since the loss of MMR (see Fig. 1 and compare with Fig. 2). Different regions from the same tumor have similar distributions and Sloci2 values.
Figure 4
Figure 4
Colorectal tumor histories inferred from their MS mutations. Each tumor history is unique, and most divisions occur before terminal clonal expansion (gray triangles). The average age is 2,300 divisions since loss of MMR. A trend toward increasing age with histologic progression is not apparent as the cancers and adenomas have similar ages. Negative clinical examinations (arrows, assuming one division per day) preceded six tumors, and for five tumors provided independent verification because they occurred before the postulated initiation of terminal clonal expansion. The negative examination in Patient I after the estimated initiation of the expansion of his Cancer-2 may indicate that it either was missed or was still too small to be detected. The data are consistent with genetic progression in the absence of visible progression, suggesting progenitor clone sizes are often below the threshold of clinical detection after loss of MMR.

Similar articles

See all similar articles

Cited by 49 PubMed Central articles

See all "Cited by" articles

Publication types

MeSH terms

Feedback