Ki67 stained tumoursCredit: Mitch Dowsett
Adhering to the Hippocratic Oath’s maxim ‘do no harm’ is not always straightforward, particularly during a global pandemic. What is a doctor to do, for instance, when a patient has operable breast cancer, but performing that surgery could increase the risk of life-threatening COVID-19?
In March 2020, as physicians worldwide were grappling with this question and hospitals were starting to delay surgeries, Mitch Dowsett, the Breast Cancer Theme Lead at The Royal Marsden Hospital and Endocrinology Team Leader at The Institute of Cancer Research, assembled an international consortium to develop a simple algorithm that could provide some guidance.
“I sat at my dining room table for about four weeks putting all the data together,” Dowsett says, reflecting on those early days. “It was remarkable just how quickly everything came together. Time was of the essence and everybody knew that.”
Dowsett’s algorithm was recently published in npj Breast Cancer, a Nature Partner Journal supported by the Breast Cancer Research Foundation. It is intended to help physicians triage postmenopausal women with ER+ HER2–breast cancers, which represent around 70% of breast cancer cases.1 During the pandemic, many within this patient group were prescribed neoadjuvant endocrine therapy (NeoET), rather than surgery, as a disease management strategy.
As a class of therapeutic, NeoET adds, blocks, or removes hormones as a means of reducing tumour burden prior to surgery. However, some ER+ HER2–tumours don’t respond to NeoET. For those patients, a delay in surgery could have a devastating impact.
That’s where the algorithm comes into play.
Members of the consortium pooled troves of unpublished data, which encompassed multiple thousands of breast cancer patients. Dowsett and his colleagues then analysed whether three biochemical markers could be used to triage patients for priority or delayed surgery.
The three markers were estrogen receptor (ER), progesterone receptor (PgR), and the proliferative marker Ki67. Both ER and PgR are routinely examined in primary breast cancer biopsies, meaning data would be readily available. Ki67 is a frequently used marker of actively proliferating cells, and has typically been used in studies examining the effect of NeoET on tumour prognosis.
Previous evidence suggested a strong correlation between tumour prognosis and Ki67 staining following multiple weeks of NeoET treatment, such that substantial Ki67 staining would indicate a poor prognosis. Despite this evidence, little had been done to connect Ki67 staining to clinical outcomes.
Dowsett and his team set out to examine tumour biopsies and medical records to identify correlations between tumour response to NeoET and the presence of ER, PgR, and Ki67 staining. With this information, they reasoned, an algorithm could be built to help triage patients based on initial biopsy results. All the team needed to do was pore through the immense amount of data provided by the consortium.
In describing the results, Dowsett says, “We realized that 60% of patients had high scores for ER and PgR and were very likely to have less than 10% Ki67 positive cells [after NeoET]. These patients could then be safely maintained on the endocrine treatment in the absence of a follow-up Ki67 test.”
Dowsett also found that 5% of patients had very low ER levels, which correlated to a poor response to NeoET. Those patients, Dowsett says, need no further analysis: “They should immediately shift to surgery or chemotherapy.”
That left 35% of patients who fell somewhere in between. This cohort showed modest ER and PgR expression, or showed ≥15% Ki67 staining on initial biopsy. Analysis of this group found that tumour response to NeoET was variable. However, analysis of Ki67 after two to four weeks of treatment could help clarify likely outcomes: After a second biopsy, patients with <10% Ki67 staining could be safely maintained on NeoET; the rest should be prioritized for surgery or chemotherapy. This simple decision tree was then translated into an easy algorithm that could help triage patients and save lives.
For Dowsett, the significance of the work is immense. “In terms of the collaboration that came together it’s been a unique experience,” he says. “I think I’ve published in excess of 700 papers, and this is one of the most rewarding ones.”