Abstract
Clonal evolution of DCIS to invasion
Ductal carcinoma in situ (DCIS) is the most common form of preinvasive breast cancer and, despite treatment, a small fraction (5-10%) of DCIS patients develop subsequent invasive breast cancer (IBC). If not treated, at least 3 out of 4 women with DCIS will not develop IBC1-3. This implies many women with non-progressive, low-risk DCIS are likely to carry the burden of overtreatment. To solve this DCIS dilemma, two fundamental questions need to be answered. The first question is, how the subsequent IBC is related to the initial DCIS lesion. The second question is how to distinguish high- from low-risk DCIS at the time of diagnosis. This is essential to take well-informed DCIS management decisions, i.e., surgery, followed by radiotherapy in case of breast conserving treatment with or without subsequent endocrine treatment, or test whether active surveillance for low-risk DCIS is safe.
How is the subsequent IBC related to the initial DCIS?
The high genomic concordance in DNA aberrations between DCIS and IBC suggest that most driver mutations and CNA events are acquired at the earliest stages of DCIS initiation. It has therefore been assumed that most solid tumours arise from a single cell and that the probability of two independent tumours arising from the same tissue is low4-6. However, lineage tracing and genomic studies strongly suggest both direct and independent clonal lineages during the initiation of DCIS and evolution to IBC. In these processes, mammary stem cells have been implicated in DCIS initiation.
Role of mammary stem cells in DCIS initiation
Lineage tracing mouse model experiments have shown the fate of individual cells and lineages that acquire mutations before a tumour is established7-9. This is also relevant for DCIS initiation10,11, as different pools of MaSCs drive the growth and development of the ductal network and are considered the cell of origin for breast cancers9,10. The ductal trees remain quiescent until puberty, during which extension, branching and termination of terminal end buds (TEBs) leads to its expansion throughout the fat pad7,12,13. Any oncogenic mutation that occurs in a fetal MaSC will spread throughout the ductal network to a large part of the ductal tree, leading to sick lobes9. By contrast, oncogenic mutations acquired by a single MaSC during puberty spread to a smaller number of offspring located in small clusters in a part of the ductal network8,14.
Direct lineage models for DCIS progression
Direct lineage models postulate that DCIS has a single cell of origin that acquires mutations and progresses to IBC15-18. This is also supported by the high genomic concordance of CNAs and mutations in synchronous DCIS–IBC regions6,15,17,19-21 and the results of a recent large longitudinal study that profiled pure DCIS and recurrent IBC using multiple sequencing techniques, which estimated direct clonal lineages in approximately ~80% of patients18.
Two distinct direct lineage models have been proposed: the evolutionary bottleneck model and the multiclonal invasion model. In the evolutionary bottleneckmodel, a single clone (or a limited number of clones) with an invasive genotype is selected and breaks through the basement membrane to migrate into surrounding tissues15,16,22, while other clones are unable to escape the ducts21-28. The multiclonal invasion model posits that most or all subclones can escape the basement membrane, establishing invasive disease6,16,17,20. The multiclonal model has not been studied widely in pure DCIS and recurrent IBC samples.
Independent lineage model for DCIS progression
DCIS lesions and IBCs can arise from different initiating cells in the same breast independently5,20,29-32. An analysis of sequential DCIS–IBC pairs in a unique, large-scale, in-depth study of 95 matched pure DCIS and recurrent IBC showed that ~20% of the IBC recurrences were indeed clonally unrelated to the primary DCIS18, as is also supported by some mathematical model studies33.
The potential role of a field effect
IBC can develop in the same breast as an initial DCIS even after treatment, which could be explained by the presence of a field effect34-37. Alternatively, the sick lobe hypothesis proposes that a single lobe harbours first-hit mutations, acquired in utero or during early mammary development37-42. This could also explain the restriction of IBC to the ipsilateral side of the breast39,43,44. Germline mutations may also explain the emergence of independent lineages in DCIS and IBC patients, lowering the threshold for cancer development32,43-46.
Convergent evolution model of DCIS progression
A third model for the emergence of IBC from DCIS is convergent evolution, in which the same mutations and CNA are selected and expanded during tumour growth such that environmental factors fuel competition between distinct clones and push them towards a similar genotype. Ultimately, two independent clonal lineages from different ancestral cells then happen to share multiple genomic aberrations or driver mutations across regions47-49. Although independent lineages are considered uncommon (~20%) in ipsilateral recurrences, they occur at much higher frequencies in contralateral recurrences (>80%), in which single-nucleotide polymorphism and comparative genomic hybridization microarrays show few (or no) genomic alterations shared in tumours from the contralateral breast cancer18,50,51.
How to distinguish high- from low-risk DCIS at the time of diagnosis?
The genomic and transcriptomic profile present at the time of DCIS diagnosis may contain crucial information on the risk of progression of DCIS to IBC. Thus far, it has been unclear whether prognostic gene expression markers can be used to separate indolent DCIS from potentially progressive DCIS. To this end, microarrays and RNA-seq have been applied for the comparison of bulk RNA from microdissected DCIS and IBC tissue. In synchronous DCIS–IBC, a limited number of transcriptional differences have been found and the few events discovered often varied extensively across different tumours52-56. Although these differences were strong, the added value of these studies is uncertain as they are often confounded by small sample size, lack of matched receptor status data, and low sample purity. Despite these limitations, these studies have implicated the epithelial-mesenchymal transition (EMT) and extracellular matrix (ECM) remodelling pathways as potentially relevant for the progression of DCIS to IBC55-62.
We studied two large DCIS cohorts: the Sloane cohort, a prospective breast screening cohort from the UK (median follow-up of 12.5 years), and a Dutch population-based cohort (NKI, median follow-up of 13 years). FFPE tissue specimens from patients with pure primary DCIS after breast-conserving surgery (BCS) +/- RT that did develop a subsequent ipsilateral event (DCIS or invasive) were considered as cases, whereas patients that did not develop any form of recurrence up to the last follow-up or death were considered as controls. We performed copy number analysis (CNA) and RNAseq analysis on 229 cases (149 IBC recurrences and 80 DCIS recurrences) and 344 controls.
We classified DCIS into the PAM50 subtypes using RNAseq data which revealed an enrichment of luminal A phenotype in DCIS that did not recur (P = 0.01, Fisher Exact test). No single copy number aberration was more common in cases compared to controls. RNAseq data did not reveal any genes significantly over/under expressed in cases versus controls after false discovery rate (FDR) correction. However, by limiting the analysis to samples that had not had RT and excluding pure DCIS recurrences we developed a penalized Cox model from RNAseq data. The model was trained on weighted samples (to correct for the biased sampling of the case control dataset) from the NKI series with double loop cross validation. Using this predicted hazard ratio, the samples were split into high, medium and low risk quantiles, with a recurrence risk of 20%, 9% and 2.5%, respectively at 5 years (p<0.001, Wald test). The NKI-trained predictor was independently validated in the Sloane No RT cohort (p = 0.02, Wald test). GSEA analysis revealed proliferation hallmarks enriched in the recurrence predictor (FDR = 0.058). The NKI-RNAseq predictor was more predictive of invasive recurrence than PAM50, clinical features (Grade, Her2 and ER) and the 12-gene Oncotype DCIS score (p < 0.001, permutation test using the Wald statistic) in both the NKI and Sloane series.
In the methylation analysis, 50 controls were compared with 35 cases. We could identify Variably Methylation Regions (VMRs) and Differentially Methylated Regions (DMRs) between cases and controls. Interestingly, VMRs were enriched in cell adhesion pathways
Conclusion
The recently acquired knowledge described above on how often the subsequent IBC is directly related to the initial DCIS and on molecular markers predicting the risk of DCIS progression is essential for accurate DCIS risk assessment. This is essential to aid accurate clinical decision making to personalize DCIS management in the near future.