NSAIDs Modulate Clonal Evolution in Barrett's Esophagus
Figure 4
Clonal evolution in individuals b and j.
(Panel A) Solid lines connect the mean amount of SGA detected across biopsies at each time point; dots correspond to individual biopsies. In individual b (black line), we observed evolutionary stasis with mean SGA remaining at 119±79 Mb over more than a decade of follow-up. In individual j (red line), a massive burst of SGA was detected in year 8.5; three years later individual j progressed to esophageal adenocarcinoma. Individual b started NSAIDs after year 5, while individual j started NSAIDs only after year 10. (Panels B and C) Circos plots showing genome-wide views of SGA over time. Each ring, labeled with a biopsy number, represents whole-genome SGA data from a different biopsy, with earlier samples toward the center. Thin black line rings separate endoscopies (time points), white background shows time periods off-NSAIDs and gray background shows time periods on-NSAIDs. Within the rings, black segments designate homozygous deletion, red single copy loss, orange copy-neutral LOH, and green copy gain. (Panel B) Circos plot of individual b. Note the appearance of “new” whole chromosome LOH at chromosome 6 and 11 in biopsy 5, taken during the off-NSAIDs period, and the detection of a minimally mutated clone in biopsies 9 and 7, taken during the on-NSAIDs period. (Panel C) Circos plot of individual j. A massive burst of SGAs was detected first in biopsy 8, in year 8.5, before the individual began regular NSAID use. Biopsy 2 (second inner ring), taken at the baseline endoscopy 8.5 years prior to the burst, shared a subset of the SGAs seen in the massively altered clone (chromosomes 10, 12, 17 and 18), and thus is likely an early example of its lineage. (Panels D and G) Consensus phylogenetic trees estimated by BEAST reveal long-term co-existence of clones. Branch lengths are scaled according to time, the tips of the phylogeny are biopsies aligned on the x-axis according to their sampling time, and all internal nodes are estimated coalescence times assuming a logistic population growth. Dashed gray line represents the start of NSAID use. (Panels E, F, H, I) Maximum parsimony trees estimated by PAUP reveal the ancestral relationships among biopsies based on shared SGA characters. Differences between the topology of the trees estimated by PAUP and BEAST are typically due to poorly supported short branches and do not affect the analyses of SGA acquisition rates. Branch lengths are scaled according to estimated number of SGAs (Panels E, H) or the amount of genome affected by SGA (Panels F, I). Note that these trees appear very different from those estimated by BEAST as the BEAST branch lengths are scaled by inferred time depth, and the rate of SGA accumulation appears highly variable with time.
