Precision targeting yields better phase I efficacy outcomes

Patients had significantly better outcomes when treatments were selected based on biomarker analysis of their tumors, in a meta-analysis of 346 phase I trials involving 351 study arms and 13,203 cancer patients.

Up to now, the purpose of phase I trials has been to determine safety based on adverse effects and tolerability. “Our analysis really shows that these days it is completely outdated and that with biomarker selection and especially genomic biomarkers, we can reach high response rates even in phase I trials,” Maria Schwaederle, Pharm.D., of the center for personalized cancer therapy at the University of California, San Diego, said in a presscast leading up to the annual meeting of the American Society of Clinical Oncology.

For the meta-analysis, Dr. Schwaederle and her colleagues performed a PubMed search on phase I clinical trials in cancer published in the years 2011-2013. They included studies using a single agent that reported adequate efficacy endpoints for response rate and progression-free survival (PSF). Overall survival was rarely reported and so was excluded.

Personalized therapy was defined as molecular biomarker–based selection of a treatment or at least half the patients in a trial having a specific tumor known to harbor the biomarker. The study compared outcomes based on a personalized strategy with those that were not.

“We think that the results were striking here,” Dr. Schwaederle said. “We can see that personalized therapy that is [based on] biomarker-selected treatments was really associated with significantly higher response rates and progression-free survival.” Multivariable analysis showed that response rates were sixfold higher with the personalized approach (30.6%, n = 58 in the personalized trials, vs. 4.9%, n = 293 in those not personalized; P less than .0001), and median PFS practically doubled (5.7 months, n = 7 in personalized trials vs. 2.95 months, n = 38 in those not personalized; P = .0002).

Most (98.3%) of the personalized arms used targeted agents. However, 76% of the arms using targeted agents did not select patients using a related biomarker and were thus nonpersonalized in their approach to treatment. A subanalysis showed that targeted drugs that were applied in a nontargeted manner had much worse response rates than targeted therapies applied using biomarker-based selection and were equivalent to cytotoxic therapies (personalized approach, 31.1%; nontargeted approach, 5.1%; P less than 0.0001; cytotoxic drugs, 4.7%, P = .63 vs. nonpersonalized approach). The median PFS was also similar for the nonpersonalized and cytotoxic strategies (3.3 vs. 2.5 months, P = .22).

Better targeting with genomic biomarkers

The investigators found that patients selected based on genomic (DNA) biomarkers had almost double the response rates (42%) of patients selected based on protein biomarkers (22.4%, P = .001). The reasons for this difference were not clear. “We might think that the target in the end is a protein, so we would expect maybe to see better results with protein expression,” Dr. Schwaederle said. But she pointed to the example of patients with lung cancer and epidermal growth factor receptor genetic alterations. If looking only for epidermal growth factor receptor protein overexpression, she said, “we wouldn’t see any response, but only the patients that have specific alterations in this gene respond very well.” So in that case, genetic detection appears to be more sensitive than a protein biomarker–based test.

Dr. Schwaederle addressed the question that has often arisen of whether targeted agents are just better therapies. “Our analysis showed that it is not just that the therapies are better but that targeted therapies must be given to the right patients,” she said. “Indeed, when targeted therapies were given to patients without a biomarker selection, the response rates were only about 5%.”

The response rate in the range of 40% using genetic biomarkers in these phase I trials suggests that incorporating such targeted approaches even at this early stage of drug testing may potentially yield useful information on efficacy.

Dr. Don Dizon, presscast moderator and chair of ASCO’s Cancer Communications Committee, said that precision medicine “is here and that we can use patient selection using either changes in a tumor’s DNA, called genomic changes, or even protein biomarkers and do much better than we have done in the past.”

Dr. Schwaederle noted that one limitation of her study is that more recent trials, unpublished at the time of her analysis, could influence the results today.

The study received funding from the Joan and Irwin Jacobs Philanthropic Fund. Two coauthors reported extensive financial relationships with pharmaceutical or other commercial and noncommercial entities.

Patients had significantly better outcomes when treatments were selected based on biomarker analysis of their tumors, in a meta-analysis of 346 phase I trials involving 351 study arms and 13,203 cancer patients.

Up to now, the purpose of phase I trials has been to determine safety based on adverse effects and tolerability. “Our analysis really shows that these days it is completely outdated and that with biomarker selection and especially genomic biomarkers, we can reach high response rates even in phase I trials,” Maria Schwaederle, Pharm.D., of the center for personalized cancer therapy at the University of California, San Diego, said in a presscast leading up to the annual meeting of the American Society of Clinical Oncology.

For the meta-analysis, Dr. Schwaederle and her colleagues performed a PubMed search on phase I clinical trials in cancer published in the years 2011-2013. They included studies using a single agent that reported adequate efficacy endpoints for response rate and progression-free survival (PSF). Overall survival was rarely reported and so was excluded.

Personalized therapy was defined as molecular biomarker–based selection of a treatment or at least half the patients in a trial having a specific tumor known to harbor the biomarker. The study compared outcomes based on a personalized strategy with those that were not.

“We think that the results were striking here,” Dr. Schwaederle said. “We can see that personalized therapy that is [based on] biomarker-selected treatments was really associated with significantly higher response rates and progression-free survival.” Multivariable analysis showed that response rates were sixfold higher with the personalized approach (30.6%, n = 58 in the personalized trials, vs. 4.9%, n = 293 in those not personalized; P less than .0001), and median PFS practically doubled (5.7 months, n = 7 in personalized trials vs. 2.95 months, n = 38 in those not personalized; P = .0002).

Most (98.3%) of the personalized arms used targeted agents. However, 76% of the arms using targeted agents did not select patients using a related biomarker and were thus nonpersonalized in their approach to treatment. A subanalysis showed that targeted drugs that were applied in a nontargeted manner had much worse response rates than targeted therapies applied using biomarker-based selection and were equivalent to cytotoxic therapies (personalized approach, 31.1%; nontargeted approach, 5.1%; P less than 0.0001; cytotoxic drugs, 4.7%, P = .63 vs. nonpersonalized approach). The median PFS was also similar for the nonpersonalized and cytotoxic strategies (3.3 vs. 2.5 months, P = .22).

Better targeting with genomic biomarkers

The investigators found that patients selected based on genomic (DNA) biomarkers had almost double the response rates (42%) of patients selected based on protein biomarkers (22.4%, P = .001). The reasons for this difference were not clear. “We might think that the target in the end is a protein, so we would expect maybe to see better results with protein expression,” Dr. Schwaederle said. But she pointed to the example of patients with lung cancer and epidermal growth factor receptor genetic alterations. If looking only for epidermal growth factor receptor protein overexpression, she said, “we wouldn’t see any response, but only the patients that have specific alterations in this gene respond very well.” So in that case, genetic detection appears to be more sensitive than a protein biomarker–based test.

Dr. Schwaederle addressed the question that has often arisen of whether targeted agents are just better therapies. “Our analysis showed that it is not just that the therapies are better but that targeted therapies must be given to the right patients,” she said. “Indeed, when targeted therapies were given to patients without a biomarker selection, the response rates were only about 5%.”

The response rate in the range of 40% using genetic biomarkers in these phase I trials suggests that incorporating such targeted approaches even at this early stage of drug testing may potentially yield useful information on efficacy.

Dr. Don Dizon, presscast moderator and chair of ASCO’s Cancer Communications Committee, said that precision medicine “is here and that we can use patient selection using either changes in a tumor’s DNA, called genomic changes, or even protein biomarkers and do much better than we have done in the past.”

Dr. Schwaederle noted that one limitation of her study is that more recent trials, unpublished at the time of her analysis, could influence the results today.

The study received funding from the Joan and Irwin Jacobs Philanthropic Fund. Two coauthors reported extensive financial relationships with pharmaceutical or other commercial and noncommercial entities.

Patients had significantly better outcomes when treatments were selected based on biomarker analysis of their tumors, in a meta-analysis of 346 phase I trials involving 351 study arms and 13,203 cancer patients.

Up to now, the purpose of phase I trials has been to determine safety based on adverse effects and tolerability. “Our analysis really shows that these days it is completely outdated and that with biomarker selection and especially genomic biomarkers, we can reach high response rates even in phase I trials,” Maria Schwaederle, Pharm.D., of the center for personalized cancer therapy at the University of California, San Diego, said in a presscast leading up to the annual meeting of the American Society of Clinical Oncology.

For the meta-analysis, Dr. Schwaederle and her colleagues performed a PubMed search on phase I clinical trials in cancer published in the years 2011-2013. They included studies using a single agent that reported adequate efficacy endpoints for response rate and progression-free survival (PSF). Overall survival was rarely reported and so was excluded.

Personalized therapy was defined as molecular biomarker–based selection of a treatment or at least half the patients in a trial having a specific tumor known to harbor the biomarker. The study compared outcomes based on a personalized strategy with those that were not.

“We think that the results were striking here,” Dr. Schwaederle said. “We can see that personalized therapy that is [based on] biomarker-selected treatments was really associated with significantly higher response rates and progression-free survival.” Multivariable analysis showed that response rates were sixfold higher with the personalized approach (30.6%, n = 58 in the personalized trials, vs. 4.9%, n = 293 in those not personalized; P less than .0001), and median PFS practically doubled (5.7 months, n = 7 in personalized trials vs. 2.95 months, n = 38 in those not personalized; P = .0002).

Most (98.3%) of the personalized arms used targeted agents. However, 76% of the arms using targeted agents did not select patients using a related biomarker and were thus nonpersonalized in their approach to treatment. A subanalysis showed that targeted drugs that were applied in a nontargeted manner had much worse response rates than targeted therapies applied using biomarker-based selection and were equivalent to cytotoxic therapies (personalized approach, 31.1%; nontargeted approach, 5.1%; P less than 0.0001; cytotoxic drugs, 4.7%, P = .63 vs. nonpersonalized approach). The median PFS was also similar for the nonpersonalized and cytotoxic strategies (3.3 vs. 2.5 months, P = .22).

Better targeting with genomic biomarkers

The investigators found that patients selected based on genomic (DNA) biomarkers had almost double the response rates (42%) of patients selected based on protein biomarkers (22.4%, P = .001). The reasons for this difference were not clear. “We might think that the target in the end is a protein, so we would expect maybe to see better results with protein expression,” Dr. Schwaederle said. But she pointed to the example of patients with lung cancer and epidermal growth factor receptor genetic alterations. If looking only for epidermal growth factor receptor protein overexpression, she said, “we wouldn’t see any response, but only the patients that have specific alterations in this gene respond very well.” So in that case, genetic detection appears to be more sensitive than a protein biomarker–based test.

Dr. Schwaederle addressed the question that has often arisen of whether targeted agents are just better therapies. “Our analysis showed that it is not just that the therapies are better but that targeted therapies must be given to the right patients,” she said. “Indeed, when targeted therapies were given to patients without a biomarker selection, the response rates were only about 5%.”

The response rate in the range of 40% using genetic biomarkers in these phase I trials suggests that incorporating such targeted approaches even at this early stage of drug testing may potentially yield useful information on efficacy.

Dr. Don Dizon, presscast moderator and chair of ASCO’s Cancer Communications Committee, said that precision medicine “is here and that we can use patient selection using either changes in a tumor’s DNA, called genomic changes, or even protein biomarkers and do much better than we have done in the past.”

Dr. Schwaederle noted that one limitation of her study is that more recent trials, unpublished at the time of her analysis, could influence the results today.

The study received funding from the Joan and Irwin Jacobs Philanthropic Fund. Two coauthors reported extensive financial relationships with pharmaceutical or other commercial and noncommercial entities.