The CARG-BC Score: Novel Tool for Predicting Chemotherapy Toxicity in Early-Stage Breast Cancer

Get Permission

As reported in the Journal of Clinical Oncology, Allison Magnuson, DO, of the University of Rochester Medical Center & Wilmot Cancer Institute, and Mina S. Sedrak, MD, MS, of the City of Hope National Medical Center, along with colleagues, have developed a novel risk tool—the Cancer and Aging Research Group–Breast Cancer (CARG-BC) score—that incorporates clinical and geriatric factors to predict the occurrence of severe toxicity in patients aged ≥ 65 years who are receiving chemotherapy for early-stage breast cancer.1

Allison Magnuson, DO

Allison Magnuson, DO

Mina S. Sedrak, MD, MS

Mina S. Sedrak, MD, MS

Study Details

In the prospective cohort study, 473 patients (from The Hurria Older PatiEnts [HOPE] with Breast Cancer Cohort Study) aged ≥ 65 years with stage I to III breast cancer from 16 U.S. institutions were evaluated for geriatric and clinical features predictive of grade 3 to 5 toxicity related to neoadjuvant or adjuvant chemotherapy. The study population consisted of 283 patients in a development cohort and 190 in a validation cohort. A risk model with weighted variable scoring was developed using logistic regression analysis, with model performance assessed by area under the receiver operating characteristic curve (AUC).

In the entire population, median age was 70 years (range = 65–86 years), 56.2% were married, 75.5% were White, 72.1% lived with someone, and 72.2% had a college education or higher. Disease stage was I in 36.2%, II in 42.9%, and III in 20.9%; 23.7% had triple-negative disease and 27.7% had HER2-positive disease. The treatment setting was adjuvant for 82.7% and neoadjuvant for 17.3%; 39.1% received an anthracycline-based regimen; 50.1% had a planned duration of treatment of more than 3 months; and 74.2% had planned primary prophylaxis with white blood cell growth factors.

The final model derived in the development cohort, termed the CARG-BC score, consisted of eight independent predictors of toxicity: planned anthracycline use (1 point), stage II or III disease (3 points), planned treatment duration of more than 3 months (4 points), abnormal liver function (normal defined as all liver tests within the normal reference ranges for each institution; 3 points), low hemoglobin (≤ 13/12 g/dL in males/females; 3 points), at least one fall in past 6 months (4 points), limited ability to walk more than 1 mile (3 points), and lack of social support (lack of someone to give good advice in a crisis; 3 points). Risk scores were calculated for each patient, and three risk groups were defined: low risk = 0 to 5 points, intermediate risk = 6 to 11 points, and high risk = at least 12 points.

Model Performance in Development and Validation Cohorts

Among 283 patients in the development cohort, chemotherapy-related grade 3 to 5 toxicities occurred in 46.3% of patients (hematologic in 25.8%, nonhematologic in 36.4%), including grade 3 in 36.0%, grade 4 in 9.9%, and grade 5 in 0.4%. Rates of grade 3 to 5 toxicity were 19% in the low-risk group (n = 93), 54% in the intermediate-risk group (n = 159), and 87% in the high-risk group (n = 30; overall P < .01). Compared with the low-risk group, odds ratios were 4.91 (95% confidence interval [CI] = 2.69–8.96, P < .001) in the intermediate-risk group and 28.13 (95% CI = 9.74–90.56, P <.001) in the high-risk group. The AUC was 0.75 (95% CI = 0.70–0.81).

Among 190 patients in the validation cohort, grade 3 to 5 toxicities occurred in 44.7% of patients (hematologic in 24.7%, nonhematologic in 35.3%), including grade 3, 4, and 5 in 36.8%, 6.8%, and 1.2%. Rates of grade 3 to 5 toxicity were 27% in the low-risk group (n = 59), 45% in the intermediate-risk group (n = 98), and 76% in the high-risk group (n = 33; overall P < 0.0001). The AUC was 0.69 (95% CI = 0.62–0.77), with a P value of .15 vs the development cohort.

Effects of Risk Level on Treatment in Total Population

Among all 473 patients in the combined cohorts, 46% developed grade 3 to 5 toxicity. Rates in low-, intermediate-, and high-risk patients were 22%, 51%, and 81% (overall P < .001). For the combined cohorts, the overall AUC for the CARG-BC score was 0.73 (95% CI = 0.68–0.77).


  • An eight-item weighted risk tool (CARG-BC score) was developed to predict chemotherapy-related grade 3 to 5 toxicity in older patients.
  • Among all patients, grade 3 to 5 toxicity occurred in 22%, 51%, and 81% of patients in low-, intermediate-, and high-risk groups, according to CARG-BC score.

Among all patients, 25% received reduced relative dose intensity (< 85% of the ideal regimen), and 23% were hospitalized during treatment. Patients in the intermediate- and high-risk groups vs the low-risk group were more likely to receive reduced relative dose intensity (29% and 45% vs 13%, overall P < .001) and to be hospitalized (27% and 38% vs 11%, overall P < .001). In addition, patients in the intermediate- and high-risk vs low-risk groups were more likely to have dose reductions (25% and 38% vs 14%, overall P < .001), dose delays (30% and 50% vs 9%, overall P < .001), and early treatment discontinuation (26% and 39% vs 13%, overall P < .001).

Comparison With Other Risk Instruments

In an analysis using data from the validation cohort, prediction of grade 3 to 5 toxicity was better with the CARG-BC score vs the generalized CARG toxicity tool (AUCs = 0.69 vs 0.56, P = .004) vs physician-rated Karnofsky performance status (AUC = 0.50, P < .001).

The investigators concluded: “[W]e developed and validated a risk score based on 8 clinical and geriatric factors that predict grade 3 to 5 chemotherapy toxicity in older adults with early-stage breast cancer. The risk score was also strongly associated with dose reductions, dose delays, reduced dose intensity, and hospitalizations. These findings may be useful to clinicians for predicting individual probability of chemotherapy toxicity and directing therapy, to researchers for designing and interpreting clinical trials, and to policymakers for allocating future resources for new strategies to mitigate the risk of chemotherapy toxicity.”

This study was conceptualized and led by the late Arti Hurria, MD, FASCO, of the City of Hope Medical Center. Dr. Hurria, a former ASCO Board member, tragically passed away at the end of this study. Dr. ­Hurria’s dream was to improve outcomes for older adults with cancer by infusing the principles of geriatrics into oncology—this study exemplifies her vision and the results bring us one step closer to realizing her dream. 

DISCLOSURE: The study was funded by the National Institute on Aging, Breast Cancer Research Foundation, Center for Cancer and Aging at City of Hope, National Cancer Institute, American Cancer Society, and Susan G. Komen for the Cure. Dr. Magnuson and Dr. Sedrak reported no conflicts of interest.


1. Magnuson A, Sedrak MS, Gross CP, et al: Development and validation of a risk tool for predicting severe toxicity in older adults receiving chemotherapy for early-stage breast cancer. J Clin Oncol. January 14, 2021 (early release online).