Breast Cancer Adjuvant Tool

Breast Cancer Adjuvant Tool by ONCOassist

The ONCOassist breast cancer adjuvant tool is based on the Predict algorithm. This was developed as a collaborative project between the Cambridge Breast Unit, the University of Cambridge Department of Oncology, and the UK’s Eastern Cancer Information and Registration Centre (ECRIC) (now part of the National Cancer Registration and Analysis Service) and was supported by an unrestricted educational grant from Pfizer Limited. (developers of ONCOassist) licensed the prediction algorithm from the relevant parties through Cambridge Enterprises. The tool is designed to give prognostication and treatment benefits to help clinicians and patients make informed decisions about treatment following breast cancer surgery. The survival estimates presented both with and without adjuvant therapy (hormone therapy, chemotherapy, and trastuzumab), are provided for 5,10, and 15 years following surgery.

Using the Predict Breast Cancer Adjuvant Tool

Use the Breast Cancer Adjuvant Tool to estimate breast cancer survival and the benefits of hormone therapy, chemotherapy and trastuzumab.

The clinical validity of a prediction model can be defined as the accuracy of the model to predict future events. The two key measures of clinical validity are calibration and discrimination.

Calibration is how well the model predicts the total number of events in a given data set. A perfectly calibrated model is one where the observed (or actual) number of events in a given patient cohort is the same as the number of events predicted by the model. Discrimination is how well the model predicts the occurrence of an event in individual patients. The discrimination statistic is a number between zero and one. It is generally obtained from the area under a receiver-operator characteristic (ROC) curve, which is a plot of the true positive rate (sensitivity) against the false positive rate (probability of false alarm).

Predict was originally validated using a dataset of over 5000 breast cancer patients from the West Midlands Cancer Intelligence Unit also diagnosed during 1999-2003 and followed for a median of 4.8 years.

We also validated Predict using a dataset from British Columbia that had been previously used for the validation of Adjuvant! Online. The British Columbia dataset included women diagnosed with breast cancer from 1989-2003 and followed for 10 years. Predict v1.0 provided overall and breast cancer-specific survival estimates that were at least as accurate as estimates from Adjuvant! The results of this validation were published in the European Journal of Surgical Oncology.

How to use Breast Cancer Adjuvant Tool?

Breast Cancer Adjuvant Tool – Model extension: HER2 status (version 1.1)

The model was updated in October 2011 to include HER2 status. Estimates for the prognostic effect of HER2 status were based on an analysis of 10,179 cases collected by the Breast Cancer Association Consortium (BCAC). A validation of the new model in the original British Columbia dataset was published in the British Journal of Cancer. This showed that the inclusion of HER2 status in the model improved the estimates of breast cancer-specific mortality, especially in HER2-positive patients.

The benefit of trastuzumab (Herceptin) is based on the estimated proportional reduction of 31 percent in the mortality rate up to five years in published trials.

Breast Cancer Adjuvant Tool – Model extension: KI67 status (version 1.2)

In v1.2, KI67 status was added to the model. The prognostic effect of KI67 was taken from published data showing that ER-positive tumours that express KI67 are associated with a 30 percent poorer relative survival.

KI67 positivity for the Predict model was defined as greater than 10 percent of tumour cells staining positive.

We have validated the version of Predict that includes KI67 using a data set from Nottingham of 1,274 women diagnosed in 1989-98 and followed for 10 years. The addition of KI67 led to a small improvement in calibration and discrimination in 1,274 patients with ER-positive disease – the area under the ROC curve improved from 0.7611 to 0.7676 (p=0.005). These data were published in BMC Cancer.

Breast Cancer Adjuvant Tool – Model re-fitting (version 2.0)

While the overall fit of Predict version 1 was good in multiple independent case series, Predict had been shown to underestimate breast cancer-specific mortality in women diagnosed under the age of 40, particularly those with ER-positive disease (See publication in Oncology Letters ). Another limitation of version 1 was the use of discrete categories for tumour size and node status which result in “step” changes in risk estimates on moving from one category to the next. For example, a woman with an 18mm or 19mm tumour will be predicted to have the same breast cancer-specific mortality if all the other prognostic factors are the same whereas breast cancer-specific mortality of women with a 19mm or 20mm tumour will differ.

In order to take into account age at diagnosis and to smooth out the survival function for tumour size and node status we refitted the Predict prognostic model using the original cohort of cases from East Anglia with follow-up extended to 31 December 2012 and including 3,787 women with 10 years of follow-up. The fit of the model was tested in the three independent data sets that had also been used to validate the original version of Predict.

Calibration in ER-negative disease validation data set: Predict v1.2 over-estimated the number of breast cancer deaths by 10 per cent (observed 447 compared to 492 predicted). This overestimation was most notable in the larger tumours and in the high-grade tumours. In contrast, the calibration of Predict v2.0 in ER-negative cases was excellent (predicted 449).

Calibration in ER-negative disease validation data set: The calibration of both Predict v1.2 and Predict v2.0 was good in ER-positive cases (observed breast cancer deaths 633 compared to 643 (v1.2) and 634 (v2.0) predicted). However, as previously described, Predict v1.2 significantly under-estimated breast cancer-specific mortality in women diagnosed with ER-positive disease at younger ages, whereas the fit of Predict v2.0 was good in all age groups.

Breast Cancer Adjuvant Tool – Model extension and correction (version 2.1)

Addition of bisphosphonates treatment option and addition of 15-year outcomes

Predict v2.0 used an inaccurate method to estimate the absolute benefit of therapy which resulted in a small overestimation of the benefits of treatment. The benefit is calculated in v2.0 as the difference in breast cancer-specific mortality with and without treatment but it is more appropriate to estimate benefit as the difference in all-cause mortality with and without treatment because, if breast cancer mortality is reduced, competing non-breast cancer mortality will increase slightly. Consequently, the overestimation of benefit was greater in older women with higher competing mortality from causes other than breast cancer. The table below shows the predicted benefits of anthracycline-based chemotherapy (2nd generation) for a woman with a 22mm, grade 2, HER2 negative, KI67 negative, clinically detected tumour with 2 positive nodes by age and ER status.

AgeEREstimated benefit at ten years (%)

The proportional reduction in the mortality rate following bisphosphonate therapy (18%) was taken from the Early Breast Cancer Trialists’ Collaborative Group (2015)(3). This is assumed to be applicable only to post-menopausal women (menopausal status is now an input in the tool).

We have extended the predictions to 15 years. While the Eastern Region Cancer Registry data used to derive the model included up to fifteen years of survival the data used for model validation only included validation of the ten-year mortality predictions. The fifteen-year mortality predictions have not been validated. We have assumed that the treatment benefits of all the treatments persist long-term with the same proportional reductions in the mortality rate from year 10 to 15 as from diagnosis to year 10. There is good evidence from some long-term follow-ups (1, 4, 5) to justify this assumption, but long-term follow-up data are not available either for trastuzumab therapy or bisphosphonates therapy.

Breast Cancer Adjuvant Tool – Future version

Addition of extended hormone therapy, radiotherapy, the addition of PR status as an input variable, and presentation of potential harms as well as benefits.

We hope to introduce the effect of progesterone receptor status (PR status) on outcomes and offer two additional treatment options: radiotherapy, and an additional five years of hormone therapy. In the longer term, we also hope to be able to display data on the recurrence of disease, showing women how long they might be able to expect without their cancer coming back. We also plan to extend the site so as to be able to display a quantification of the potential harms of treatments (i.e. the proportion of similar women expected to suffer each potential side effect or adverse event). This will enable the potential harms to be considered alongside the potential benefits of each treatment.

Other Adjuvant Tools we offer:

  • Colon Tool – Prediction algorithm estimating survival rates for colon cancer based on risk factors and treatment.
  • GIST Tool – Prediction algorithm estimating relapse rates for GIST.
  • Lung Cancer – Prediction algorithm estimating survival rates for lung cancer based on risk factors and treatment.
  • Melanoma Tool – This tool is designed to show the efficacy outcome with adjuvant BRAF targeted and immunotherapy options in recent clinical trials for stage III melanoma patients.