Nomogram predicted risk for locoregional recurrence

SAN FRANCISCO — A nomogram that accounts for clinicopathologic features predicted risk for locoregional recurrence among women who underwent accelerated partial-breast irradiation for early-stage breast cancer, according to results of a multi-institutional study presented at the Breast Cancer Symposium.

“This practical, easy-to-use nomogram offers a potential guide for appropriate selection of patients at low risk for locoregional recurrence,” Jessica L. Wobb, MD, senior resident in radiation oncology at Beaumont Health System in Birmingham, Mich., said during a presentation. “It also can aid in determining appropriate candidates for off-protocol utilization of accelerated partial-breast irradiation. It could be in conjunction with consensus guidelines from different societies to make a more informed, more personalized decisions for patients.”

Few tools exist to help guide clinicians with regard to locoregional recurrence risk among patients who want to undergo accelerated partial-breast irradiation (APBI).

Wobb and colleagues initially evaluated 2,000 cases, 551 (27.5%) of which involved women treated at William Beaumont Hospital. The other 1,449 (72.5%) involved women treated as part of the ASBrS MammoSite Registry Trial.

In the majority of cases (n=1,689; 84.8%), women underwent APBI with balloon-based brachytherapy. Other methods included 3D conformal radiotherapy (n=213; 10.7%) and multiplanar interstitial catheters (n=98; 4.9%).

Wobb and colleagues excluded more than 400 cases due to incomplete data, so the final analysis included 1,573 cases. The median age was 65 years (range, 51-94). The majority of patients were older than age 50 years and post-menopausal. The majority of patients had tumors less than 20 mm in size, T1N0 disease, negative margins, ER-positive status, and tumor grade of 2 to 3.

Researchers prospectively gathered clinicopathologic variables, and they used the Cox proportional hazards regression model to generate risk estimates and hazard ratios. They generated a bias-corrected index to validate the nomogram, then used a concordance index (C-index) for cross-validation.

Median follow-up was 5.5 years (range, 0.9-18.3).

Overall, the rate of local failure among the entire cohort was 4%.

Results of univariate analysis showed five variables — age <50 years, close or positive margins (greater than 0.1 mm but less than 2 mm), pre- or perimenopausal status, high tumor grade and ER negativity — were associated with higher rates of locoregional recurrence.

Wobb and colleagues then created a multivariable model with adjusted estimates from those five independent covariates.

The two predictors of local recurrence that carried the highest HRs were ER negativity (HR=2.8 vs. ER-positive status) and high tumor grade (HR=1.9 vs. low tumor grade). ER negativity is taken into consideration in current ASTRO consensus guidelines but tumor grade is not.

Although age, menopausal status and margins did not reach statistical significance, they were added to the model because they did improve the fit of the overall model, Wobb said.

“These five independent covariates were used to create adjusted estimates, weighting each on a scale of 0 to 100,” the researchers wrote. “The total score is identified on a points scale to obtain the probability of locoregional recurrence over the study period.”

Researchers calculated a C-index of 0.641, demonstrating its strong concordance for the prediction of locoregional recurrence.

Wobb applied the model to two cases during her presentation.

The first case involved a 63-year-old postmenopausal patient with negative margins, and ER-positive, grade 1 disease. The nomogram produced a total score of 0 points for this case, which translates to a predictive value of less than 5% chance for locoregional recurrence within 5 years.

The second case involved a 50-year-old postmenopausal woman with positive margins, and ER-positive, grade 2 disease.

The weight of positive margins and a higher tumor grade in the nomogram yielded a total point score of 103, which translates to a 16% predicted locoregional recurrence rate at 5 years. However, if the patient were taken back to the OR and re-resected with negative margins, her score would drop to a point where it would halve her risk for locoregional recurrence to about 7.5%.

For more information:

Wobb JL. Abstract #59. Presented at: Breast Cancer Symposium; Sept. 4-6, 2014; San Francisco.

Disclosure: One of the researchers reports an employment relationship with 21st Century Oncology.