A statistical model that combines multiple variables such as tumor size and lymph-node status into a single score can improve patient selection for breast cancer clinical trials and produce stronger results.
Lajos Pusztai, MD, DPhil
Investigators at Yale Cancer Center believe that a statistical model that combines multiple variables such as tumor size and lymph-node status into a single score can improve patient selection for breast cancer clinical trials and produce stronger results.
Clinical trials often use selection methods that underestimate how many patients can be treated successfully with standard-of-care therapies, and won't benefit from a more effective experimental drug. Many factors contribute to a breast cancer patient’s risks of recurrence or death, including tumor size and overall health. The problem, Lajos Pusztai, MD, DPhil, chief of Breast Medical Oncology at Yale Cancer and the study’s senior corresponding author, said in a press release, is that clinical trials have historically selected patients using mostly tumor size and lymph node status.
The result is that patients included in the control arm of a trial who receive standard-of-care treatment experience too few cancer recurrences or deaths to allow statistically conclusive comparisons with patients receiving experimental therapy.
“If there are not enough events, one can miss a truly effective drug that could help patients who remain at risk for recurrence despite current best therapies,” Pusztai said.
To combat that problem, Pusztai et al simulated 2-arm, 1:1 randomized clinical trials with a target hazard ratio (HR) of 0.70 for recurrence-free survival under 4 different accrual scenarios. In scenario 1, the proportion of patients in each tumor size and clinical nodal status cohort and their corresponding 5-year survival rates came from the MD Anderson Cancer Center Department of Breast Medical Oncology database.
Scenarios 2 and 3 were created by increasing the proportion of low-risk patients (T2/N0) from the observed 40% to 55% and 70%, respectively, and correspondingly lowering the proportion of high-risk (T2/N1, T3/N1, T4) patients. Scenario 4 represented increasing both the proportion and the 5-year survival of the T2/N0 cohort. The clinical trial power of each scenario was determined based on 5000 simulated trials.
Investigators calculated baseline prognostic risk, defined as the risk for recurrence with surgery alone, and residual risk, defined as the risk for recurrence that remains after completion of all planned adjuvant therapies, using Adjuvant! Online, a widely used risk prediction model.
The model combined variables such as tumor size, nodal status, patient age, overall health, and benefit from standard-of-care therapy into a single score corresponding to the predicted risk for recurrence after receiving current best therapies.
Patients in the simulations with a <10% risk for recurrence were classified as low risk. Those with a 10% to 20% risk were classified as intermediate and patients with a 20% risk were considered high risk.
Investigators found that the baseline and the treatment-adjusted residual risks differed substantially. Among the 443 consecutive patients, 26% had intermediate baseline risk and 74% had high baseline risk for recurrence. After adjusting for adjuvant therapy effect, 28% were low risk, 45% were intermediate, and 27% had high residual risk.
Investigators then examined residual risk distribution in 3 patient cohorts corresponding to patients who would be eligible for 3 ongoing adjuvant randomized clinical trials. The interquartile ranges ranged from 15 to 31, suggesting that the power of these trials could vary unpredictably based on the residual risk distribution of the accrued population.
With a sample size of 800, the respective power associated with scenario 1 was 0.87, compared with 0.84 in scenario 2, 0.80 in scenario 3, and 0.76 in scenario 4. Investigators emphasized these changes in power occurred even though all patients met eligibility criteria based on tumor size and nodal status.
However, the power of the trial increases when eligibility is defined as >40% residual risk. An 800-patient randomized clinical trial retains a power of 82% even when 70% of patients have T2/N0 disease.
In scenario 1, a trial would have an 82% power to detect an HR of 0.70 with a sample size of 600 if eligibility is defined as a minimum 10-year residual risk of 50%, whereas a trial with the same sample size using combinations of nodal size and tumor size for eligibility would have a power of 0.77.
“This approach guarantees that the statistical power of the study is adequate to demonstrate if the new treatment is really effective or not. We also can expose fewer patients to the side effects of the new treatment. That's good for the patients,” Pusztai said. “We can select patients who really require clinical trials because their outcome with current treatments is less than optimal.”
Wei W, Kurita T, Hess KR, et al. Comparison of residual risk—based eligibility vs tumor size and nodal status for power estimates in adjuvant trials of breast cancer therapies [published online January 25, 2018]. JAMA Oncol. doi:10.1001/jamaoncol.2017.5092.
In scenarios 2 and 3, the power of the trial drops as the proportion of T2/N0 patients increases and the proportion of higher-risk cohorts decreases. Investigators said the trial becomes even weaker if the 5-year survival of the T2/N0 cohort increases as in scenario 4, which might happen if the standard of care improves over time.