ľֱ

Study Validates Accuracy of PREDICT Prognostication Tool in Breast Cancer

— Findings suggest it's a suitable decision aid for U.S. oncologists and patients

MedpageToday
Photo of a doctor taking notes from the PREDICT Tool on his computer

The prognostication tool was able to estimate overall survival (OS) with decent accuracy in a large independent U.S. dataset of breast cancer patients, researchers reported.

From more than 700,000 patients in the National Cancer Database (NCDB), the free online tool estimated median and mean 5-year OS rates ranging from 83.3% to 84.4%, as compared with an observed rate of 89.7%, found Nickolas Stabellini, BS, of Case Western Reserve University School of Medicine in Cleveland, and colleagues.

Estimated 10-year OS rates with the U.K-developed tool ranged from 69.4% to 73.8% versus the observed rate of 78.7%, they wrote in the .

PREDICT's performance was measured by the area under the curve (AUC) of time-dependent receiver operating characteristic curves, which translated into values of 0.78 and 0.76 for survival at 5 and 10 years, respectively.

"[W]e have shown that PREDICT is a clinically useful tool that can be routinely used by medical oncologists as part of their decision-making and estimation of clinical risk and the impact of breast cancer adjuvant therapies on OS," they wrote.

In an , Paul D.P. Pharoah, PhD, of Cedars-Sinai Medical Center in Los Angeles, noted that no prediction model is perfect, "and PREDICT is no exception."

"[B]ut the calibration and discrimination were within acceptable limits," he added. "The validation of the performance of the model using a large cohort of patients from the United States confirms it is a suitable decision aid for patients and oncologists in the United States."

PREDICT is a widely used web-based tool that estimates the absolute risk of dying from breast cancer and the absolute benefit of adjuvant therapies on OS in patients with breast cancer. The authors noted that while PREDICT has been validated in other countries and populations, this is the first study to validate it in a large U.S.-based dataset of patients with breast cancer.

Study authors looked at 708,652 breast cancer patients in the NCDB diagnosed with primary unilateral invasive disease from 2004 to 2012 and with at least 5 years of follow-up.

Women had a median age of 58 years and were mostly white (85.4%) and non-Hispanic (88.4%). A majority had private health insurance (59.2%), and most had low comorbidity scores (85.9%), were from metropolitan areas (86.4%), and were diagnosed with estrogen receptor (ER)-positive breast cancer (79.6%).

About half of the patients received adjuvant chemotherapy, two-thirds received adjuvant endocrine therapy, 60% underwent a partial mastectomy, and 59% had one to five axillary sentinel nodes removed. Median follow-up time was 97.7 months.

PREDICT performed best in Black patients, non-Hispanic patients, those with stage I breast cancer, those with ER-positive disease, those with HER2-negative disease, and patients receiving adjuvant endocrine therapy only.

The model performed worst in patients with high Charlson-Deyo comorbidity scores, where PREDICT overestimated survival.

The authors acknowledged their study had several limitations. For example, they noted that the NCDB doesn't provide detailed treatment data regarding the type or duration of either adjuvant chemotherapy or endocrine therapy. Plus, the study excluded patients who received neoadjuvant therapy -- the standard of care for most cases involving triple-negative or HER2-positive breast cancer.

In his commentary, Pharoah said that another important limitation was that the model's performance is based on all-cause mortality rather than breast cancer-specific mortality.

"The key output of PREDICT is the absolute benefit of systemic therapies, and this is based on the effect of therapy on breast cancer-specific mortality," he wrote. "Consequently, it is important that PREDICT should provide an accurate estimate of breast cancer-specific mortality."

"It seems likely that if the prediction of all-cause mortality is reasonable, then the breast cancer-specific mortality will also be adequate, but this assumption cannot be tested with these data," said Pharoah.

  • author['full_name']

    Mike Bassett is a staff writer focusing on oncology and hematology. He is based in Massachusetts.

Disclosures

Stabellini reported support from the Sociedade Beneficente Israelita Brasileira Albert Einstein, co-authors reported no disclosures.

Pharoah reported receiving commercial licensing fees from Cambridge Enterprises.

Primary Source

Journal of the National Comprehensive Cancer Network

Stabellini N, et al "Validation of the PREDICT prognostication tool in US patients with breast cancer" J Natl Compr Canc Netw 2023; DOI:10.6004/jnccn.2023.7048.

Secondary Source

Journal of the National Comprehensive Cancer Network

Pharoah PDP "Discussing validation of the PREDICT prognostication tool in patients with breast cancer" J Natl Compr Canc Netw 2023; DOI:10.6004/jnccn.2023.7088.