User login
SAN ANTONIO – PARPi 7, BRCAness, and MammaPrint high1/(ultra)high2 signatures could help predict response to combination therapy with the poly ADP ribose polymerase (PARP) inhibitors veliparib and carboplatin among high-risk breast cancer patients and thereby improve patient care, according to findings from the I-SPY 2 clinical trial.
“I-SPY 2 is a phase II adaptively randomized neoadjuvant clinical trial with a shared control arm where patients receive standard neoadjuvant chemotherapy, and up to 4 simultaneous investigational arms. The primary endpoint of the trial is pathologic complete response, or pCR. The goal is to match therapies with the most responsive breast cancer subtypes,” Denise M. Wolf, PhD, of the University of California, San Francisco, explained at the San Antonio Breast Cancer Symposium.
The current analysis of I-Spy 2 data focuses on veliparib/carboplatin (VC), and the subtypes Dr. Wolf mentioned are defined by hormone receptor (HR) and HER2 expression, and by MammaPrint high1 or (ultra)high2 risk status, which, “roughly speaking, is a further stratification of the poor prognosis group into high- and extra-high-risk groups,” she said.
“This arm was open to HER2-negative patients, and graduated successfully in the triple-negative subset,” she noted, explaining that “agents or combinations graduate for efficacy if they reach 85% predicted probability of success in a subsequent phase III trial in the most responsive patient subset.”
The biomarker component of the trial aimed to evaluate biomarkers associated with the mechanisms of action of each investigational agent along with the predefined subsets.
Although the findings require verification in a larger trial, Dr. Wolf and her colleagues found that three biomarkers – the PARPi 7 gene signature, the 77-gene BRCAness signature (which both relate to DNA damage repair deficiency), and the MammaPrint high1 and (ultra)high2 risk categories – were each moderately correlated with treatment response in 72 patients receiving veliparib/carboplatin (VC), but not in 44 controls, and that the treatment-biomarker interactions retained statistical significance after adjusting for hormone receptor status.
“And so we asked the question, ‘Are these signatures identifying the same patients – and if not, might there be a way to combine them to identify a subset of patients who are especially likely to respond to VC?’” she said.
Further analysis showed that even though each of the biomarkers was a predictor of response, the biomarkers did not appear to identify exactly the same patients, therefore combining them might be of benefit.
“We did this using a simple voting scheme to combine pairs of biomarkers,” she said, adding that if the two paired biomarkers predicted resistance, the biomarker also predicted resistance. If only 1 predicted resistance, the combination predicted resistance, and only if both biomarkers predicted sensitivity did the combination predict sensitivity.
In the graduated triple-negative subset, for example, the 40% of patients who were PARPi 7–high and MammaPrint high2 (the two biomarkers most predictive of response) a dramatic separation was seen in the pCR probability curves, with an estimated pCR of 79% with VC treatment vs. 23% in the control arm.
“In contrast, triple-negative patients who were negative for one or more of the sensitivity markers had nearly overlapping probability response curves, from which we conclude that nearly all of the specific sensitivity to VC seen in the triple-negative patients is found in that subset who are positive for both sensitivity markers,” she said.
Additionally, although only 9% of HR-positive/HER2-negative patients were PARPi 7-high and MammaPrint high2, those patients also appeared to be more responsive to VC vs. the control arm (49% vs. 15%).
The results also demonstrate the value of an exploratory voting method for combining multiple biomarkers for the same treatment, Dr. Wolf noted.
However, the findings are limited by the small sample size and need to be evaluated in larger trials, and the biomarkers also need to be evaluated in carboplatin-only arms in order to “tease out whether the biomarkers are really specific to a PARP inhibitor with carboplatin or whether they might also apply to carboplatin alone.
“Our ongoing and future work is to develop predictive biomarkers for other I-Spy 2 agents,” she said.
Dr. Wolf reported having no disclosures.
SAN ANTONIO – PARPi 7, BRCAness, and MammaPrint high1/(ultra)high2 signatures could help predict response to combination therapy with the poly ADP ribose polymerase (PARP) inhibitors veliparib and carboplatin among high-risk breast cancer patients and thereby improve patient care, according to findings from the I-SPY 2 clinical trial.
“I-SPY 2 is a phase II adaptively randomized neoadjuvant clinical trial with a shared control arm where patients receive standard neoadjuvant chemotherapy, and up to 4 simultaneous investigational arms. The primary endpoint of the trial is pathologic complete response, or pCR. The goal is to match therapies with the most responsive breast cancer subtypes,” Denise M. Wolf, PhD, of the University of California, San Francisco, explained at the San Antonio Breast Cancer Symposium.
The current analysis of I-Spy 2 data focuses on veliparib/carboplatin (VC), and the subtypes Dr. Wolf mentioned are defined by hormone receptor (HR) and HER2 expression, and by MammaPrint high1 or (ultra)high2 risk status, which, “roughly speaking, is a further stratification of the poor prognosis group into high- and extra-high-risk groups,” she said.
“This arm was open to HER2-negative patients, and graduated successfully in the triple-negative subset,” she noted, explaining that “agents or combinations graduate for efficacy if they reach 85% predicted probability of success in a subsequent phase III trial in the most responsive patient subset.”
The biomarker component of the trial aimed to evaluate biomarkers associated with the mechanisms of action of each investigational agent along with the predefined subsets.
Although the findings require verification in a larger trial, Dr. Wolf and her colleagues found that three biomarkers – the PARPi 7 gene signature, the 77-gene BRCAness signature (which both relate to DNA damage repair deficiency), and the MammaPrint high1 and (ultra)high2 risk categories – were each moderately correlated with treatment response in 72 patients receiving veliparib/carboplatin (VC), but not in 44 controls, and that the treatment-biomarker interactions retained statistical significance after adjusting for hormone receptor status.
“And so we asked the question, ‘Are these signatures identifying the same patients – and if not, might there be a way to combine them to identify a subset of patients who are especially likely to respond to VC?’” she said.
Further analysis showed that even though each of the biomarkers was a predictor of response, the biomarkers did not appear to identify exactly the same patients, therefore combining them might be of benefit.
“We did this using a simple voting scheme to combine pairs of biomarkers,” she said, adding that if the two paired biomarkers predicted resistance, the biomarker also predicted resistance. If only 1 predicted resistance, the combination predicted resistance, and only if both biomarkers predicted sensitivity did the combination predict sensitivity.
In the graduated triple-negative subset, for example, the 40% of patients who were PARPi 7–high and MammaPrint high2 (the two biomarkers most predictive of response) a dramatic separation was seen in the pCR probability curves, with an estimated pCR of 79% with VC treatment vs. 23% in the control arm.
“In contrast, triple-negative patients who were negative for one or more of the sensitivity markers had nearly overlapping probability response curves, from which we conclude that nearly all of the specific sensitivity to VC seen in the triple-negative patients is found in that subset who are positive for both sensitivity markers,” she said.
Additionally, although only 9% of HR-positive/HER2-negative patients were PARPi 7-high and MammaPrint high2, those patients also appeared to be more responsive to VC vs. the control arm (49% vs. 15%).
The results also demonstrate the value of an exploratory voting method for combining multiple biomarkers for the same treatment, Dr. Wolf noted.
However, the findings are limited by the small sample size and need to be evaluated in larger trials, and the biomarkers also need to be evaluated in carboplatin-only arms in order to “tease out whether the biomarkers are really specific to a PARP inhibitor with carboplatin or whether they might also apply to carboplatin alone.
“Our ongoing and future work is to develop predictive biomarkers for other I-Spy 2 agents,” she said.
Dr. Wolf reported having no disclosures.
SAN ANTONIO – PARPi 7, BRCAness, and MammaPrint high1/(ultra)high2 signatures could help predict response to combination therapy with the poly ADP ribose polymerase (PARP) inhibitors veliparib and carboplatin among high-risk breast cancer patients and thereby improve patient care, according to findings from the I-SPY 2 clinical trial.
“I-SPY 2 is a phase II adaptively randomized neoadjuvant clinical trial with a shared control arm where patients receive standard neoadjuvant chemotherapy, and up to 4 simultaneous investigational arms. The primary endpoint of the trial is pathologic complete response, or pCR. The goal is to match therapies with the most responsive breast cancer subtypes,” Denise M. Wolf, PhD, of the University of California, San Francisco, explained at the San Antonio Breast Cancer Symposium.
The current analysis of I-Spy 2 data focuses on veliparib/carboplatin (VC), and the subtypes Dr. Wolf mentioned are defined by hormone receptor (HR) and HER2 expression, and by MammaPrint high1 or (ultra)high2 risk status, which, “roughly speaking, is a further stratification of the poor prognosis group into high- and extra-high-risk groups,” she said.
“This arm was open to HER2-negative patients, and graduated successfully in the triple-negative subset,” she noted, explaining that “agents or combinations graduate for efficacy if they reach 85% predicted probability of success in a subsequent phase III trial in the most responsive patient subset.”
The biomarker component of the trial aimed to evaluate biomarkers associated with the mechanisms of action of each investigational agent along with the predefined subsets.
Although the findings require verification in a larger trial, Dr. Wolf and her colleagues found that three biomarkers – the PARPi 7 gene signature, the 77-gene BRCAness signature (which both relate to DNA damage repair deficiency), and the MammaPrint high1 and (ultra)high2 risk categories – were each moderately correlated with treatment response in 72 patients receiving veliparib/carboplatin (VC), but not in 44 controls, and that the treatment-biomarker interactions retained statistical significance after adjusting for hormone receptor status.
“And so we asked the question, ‘Are these signatures identifying the same patients – and if not, might there be a way to combine them to identify a subset of patients who are especially likely to respond to VC?’” she said.
Further analysis showed that even though each of the biomarkers was a predictor of response, the biomarkers did not appear to identify exactly the same patients, therefore combining them might be of benefit.
“We did this using a simple voting scheme to combine pairs of biomarkers,” she said, adding that if the two paired biomarkers predicted resistance, the biomarker also predicted resistance. If only 1 predicted resistance, the combination predicted resistance, and only if both biomarkers predicted sensitivity did the combination predict sensitivity.
In the graduated triple-negative subset, for example, the 40% of patients who were PARPi 7–high and MammaPrint high2 (the two biomarkers most predictive of response) a dramatic separation was seen in the pCR probability curves, with an estimated pCR of 79% with VC treatment vs. 23% in the control arm.
“In contrast, triple-negative patients who were negative for one or more of the sensitivity markers had nearly overlapping probability response curves, from which we conclude that nearly all of the specific sensitivity to VC seen in the triple-negative patients is found in that subset who are positive for both sensitivity markers,” she said.
Additionally, although only 9% of HR-positive/HER2-negative patients were PARPi 7-high and MammaPrint high2, those patients also appeared to be more responsive to VC vs. the control arm (49% vs. 15%).
The results also demonstrate the value of an exploratory voting method for combining multiple biomarkers for the same treatment, Dr. Wolf noted.
However, the findings are limited by the small sample size and need to be evaluated in larger trials, and the biomarkers also need to be evaluated in carboplatin-only arms in order to “tease out whether the biomarkers are really specific to a PARP inhibitor with carboplatin or whether they might also apply to carboplatin alone.
“Our ongoing and future work is to develop predictive biomarkers for other I-Spy 2 agents,” she said.
Dr. Wolf reported having no disclosures.
AT SABCS 2016
Key clinical point:
Major finding: Estimated pCR in PARPi 7–high and MammaPrint 2 triple-negative patients was 79% vs. 23% for the VC arm vs. control arm.
Data source: The phase II adaptively randomized I-Spy 2 clinical trial of 116 subjects.
Disclosures: Dr. Wolf reported having no disclosures.