NEW YORK (360Dx) – Researchers from The Royal Marsden NHS Foundation Trust and Queen Mary University of London have developed an online prognostic tool that uses standard clinical, pathological variables collected from breast cancer patients at the point of diagnosis to predict the risk of their cancers recurring more than five-years after diagnosis and treatment.
The so-called Clinical Treatment Score post-5-years tool, or CTS5, is described in a recent paper published in the Journal of Clinical Oncology. According to its developers, it offers an easy way for clinicians to group patients based on their recurrence risk after completing the standard five years of endocrine therapy. More specifically, it lets clinicians identify which patients are at high enough risk of their cancers returning to benefit from additional years of treatment beyond the standard five years recommendation. They can also identify patients with a low risk of recurrence and no need for additional therapy.
Jack Cuzick, director of the Wolfson Institute of Preventive Medicine at Queen Mary University of London and one of the lead researchers on the paper and CTS5's development, said in an interview that the current iteration of the tool builds on an existing model called the Clinical Treatment Score or CTS which he and his colleagues developed previously. CTS offered insights into recurrence risk within the first five years after the standard treatment regimen – surgery, chemotherapy, and/or radiation therapy – is completed.
According to one estimate mentioned in the paper, approximately 80 percent of patients are diagnosed as estrogen receptor-positive breast cancer cases. In almost all cases, these patients are prescribed five years of additional hormone therapy after undergoing the standard treatment regimen to lower the risk of their cancers returning.
Existing studies show that the continued treatment reduces patient mortality by approximately 30 percent with five years of treatment using tamoxifen and by 40 percent with five years of treatment using aromatase inhibitors. However, the dynamics of the first five years posttreatment and those of the second five years are "quite separate issues," Cuzick said. Since cancers can recur occur after the five-year treatment regimen ends, one of the real challenges for clinicians is deciding whether to continue treatment for an additional five years.
But hormone therapy can have major side effects including effects on bone tissue and menopausal symptoms. Because hormone-sensitive breast cancer is one of the few cancers where late recurrence is common, being able to predict who is at high risk is critical to making decisions about who is likely to benefit from extended treatment.
Cuzick expects that making these sorts of determinations will be the main value that CTS5 brings to the oncology space. Currently, there are no calculators available that are designed to predict late recurrence based purely on clinical information, he said. Some commercial test options include scores for measuring recurrence risk based on multigene expression profiles. The shortlist includes the Oncotype Dx recurrence score offered as part of the Genomic Health's tests, the PAM50-based Prosigna recurrence risk score marketed by NanoString, and the EndoPredict recurrence risk score offered Myriad Genetics.
Though these tests offer vital information at the point of diagnosis, they are not calibrated for application five years after diagnosis, the researchers claim in the paper. Moreover, since CTS5 does not require molecular markers to make risk predictions, it can be used in cases where patients have not undergone testing.
"If you had to choose between, say a molecular test or CTS5," Cuzick said, "CTS5 provides more information about late recurrence than the molecular test." However, he noted that molecular testing and CTS5 are independent sources of information so combining them offers the most prognostic value. A study done in 2016 comparing a score based on clinical features and five molecular tests including some of those mentioned above came to a similar conclusion. Researchers in that study concluded that molecular prognostic signatures should not be looked at in isolation and that incorporating clinical information would improve their prognostic power.
Designed with clinicians in mind, CTS5 uses information that is routinely collected at the point of diagnosis. This includes node status, tumor size and grade, as well as the patient's age. "We basically took [the CTS] model that we developed … and modified it … to look at how the original data that you would collect at the time of diagnosis would predict what's going to happen after five years," Cuzick said.
The researchers developed their algorithm using data from two previously published studies that included detailed clinical follow-up for 10 years. These were the Arimidex, Tamoxifen, Alone or in Combination (ATAC) and BIG 1-98 studies. Together, these provide information on over 11,000 postmenopausal women with ER-positive breast cancer who received five years of hormone therapy using tamoxifen, anastrozole, or letrozole.
Using the ATAC study, which includes data from 4,735 postmenopausal patients, the researchers measured how many women developed metastasis five to 10 years post-hormone therapy. They then combined these results with clinical information collected at the point of diagnosis to generate the risk equation that underlies CTS5. The researchers tested the equation using data from the BIG I-98 study which comprises 6,700 postmenopausal patients. Their results showed the CTS5 was able to accurately categorize women as low, intermediate, or high risk of recurring breast cancer after five years of hormone therapy.
Specifically, according to results reported in the paper, when CTS5 was applied to the training dataset, 42 percent of the women were categorized as low risk, about 31 percent were categorized as intermediate risk, and about 28 percent were categorized as high risk for late distant recurrence. A similar breakdown was observed in the validation dataset.
CTS5 is free for anyone that wants to use it and there are no immediate plans to commercialize it. However, commercial firms who want to use the solution as part of their diagnostic test offerings will need to license it from the developers for a yet-to-be named fee. So far, clinicians at the Royal Marsden are using the calculator but Cuzick expects that its use will become more widespread in future. "We purposely made it very simple so that any clinician can use it, put in the things that they understand and get an answer which helps guide them," he said.
Moving forward, the researchers are exploring the possibility of adding functionality that will enable clinicians to incorporate information about molecular markers if available, Cuzick said. They are also working on a predictive model for predicting risk in the context of prostate cancer cases and they might develop one for use in colorectal cancer as well.
Cuzick and his colleagues note in the paper that CTS5 has been validated for use in ER-positive HER2 negative breast cancer cases. As such, clinicians treating patients with trastuzumab, which is indicated for HER2-positive breast cancer cases, should exercise caution when using CTS5 predictions as the tool has not been validated for this group.