We are all familiar with the diagram which features a sequence of cross-sections of a breast duct as it progresses from normal epithelium through to atypical hyperplasia, ductal carcinoma in situ (DCIS), DCIS with microinvasion to invasive breast cancer, ultimately leading to metastatic breast cancer. Although we can quibble about the details, this model leads us to infer that if we can eradicate the lesion at the DCIS stage, we can avoid the downstream consequences. Recently, the proposed natural history of early breast cancer is challenged by the early findings of the COMET Study [1]. In this study, the research team randomised 957 women with low-grade DCIS to surgical ablation up front (conventional care) or to ‘watchful waiting”. In the latter arm, patients only underwent surgery if the DCIS grew in size or progressed to invasive cancer. The hope was that many women with a low-risk form of DCIS might be spared surgery without experiencing any adverse consequences. The early results were presented after only two years of follow up. In the conventional care group (immediate surgery), 5.9% of the women progressed to invasive cancer, compared to 4.2% in the watchful-waiting arm. But there was a lot of cross-over; many women did not comply with their assignment. Of the 246 women who underwent surgery, the two year risk of invasive cancer was 8.7%, versus 3.1% for the 427 women who went with watchful-waiting. The authors consider this a positive result “supporting the conclusion that active monitoring is superior to guideline-concordant care” We might consider this result to be an example of ‘de-escalation’, that is, reducing the extent of surgery did not result in worse outcomes. Perhaps the follow-up time was too short. But I think this study tells us more. If we remove the source of the invasive cancer, then why didn’t we reduce the risk of invasive cancer later on? That is what the diagram says, no? I think this observation challenges conventional wisdom and prompts us to consider the possibility that the DCIS cells are not the source of invasive cancer. Perhaps both phenomena are the results of common precursor, rather than a linear step-wise progression. Is there a hidden underlying precursor which leads to both DCIS and invasive cancer?

This should not come as surprising. We have come a long way since Halsted’s radical mastectomy. Yet I am not aware of any study—randomized or observational—where a surgical intervention in breast cancer treatment led to a clear reduction in mortality. We applauded Veronesi and Fisher who showed that replacing the Halsted radical mastectomy with the much less invasive quadrantectomy [2] or lumpectomy [3] did not increase mortality. Two classical examples of de-escalation. Fisher also showed that removing positive lymph nodes behind didn’t improve survival [3]. More recently, we have shown that we could prevent invasive ipsilateral cancer in DCIS patients with mastectomy (versus lumpectomy with or without radiotherapy) but this did not lead to a decrease in metastatic events [4]. Bilateral mastectomy prevents contralateral cancer but does not reduce mortality [5]. If the outcomes of two protocols are the same then either they are equally effective or neither of them works. Surgeons have coined the term ‘de-escalation’ meaning less surgery, leading to a better quality of life without jeopardizing survival. I think we must consider the parsimonious alternative—that is, if the two outcomes are the same, neither of them is effective. I am all in favor of de-escalation, but I think we should keep our eyes on the prize which is preventing women from dying of breast cancer. Switching to the population scale, how is that that the huge increase in the incidence of DCIS in the last three decades is not offset by a decrease in the incidence of stage IV breast cancer? This leads me to question the accuracy of canonical diagram of breast cancer (which I still present to my medical students). Perhaps the conventional model is wrong and we must draw a new figure to explain cancer progression.

Thirty years on, I have come to believe that the various manifestations of breast cancer are the result of a common precursor rather than a linear succession. Thankfully, I am not alone. Some have called this the parallel model of breast cancer progression [6,7,8]. I think it is time we devote our research efforts to identifying the culprit precursors that are the underlying cause of breast cancer in all its manifestations.