Gastrointestinal Cancers Symposium 2012

SAN FRANCISCO – Despite disappointing results in an international randomized phase III trial, investigators are hoping a biomarker will help them to find a role for everolimus in advanced gastric cancer.

The oral mammalian target of rapamycin (mTOR) inhibitor appeared to have limited efficacy as second- or third-line therapy in the GRANITE-1 study of more than 650 patients.

Overall survival was statistically indistinguishable between patients given everolimus (Afinitor) and those given a placebo, each added to best supportive care. Statistically, progression-free survival was significantly better with everolimus, but the absolute benefit was small, just 0.3 months.

"This trial was negative," Dr. Eric Van Cutsem told attendees when he reported the results at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology. But the story did not end there.

"What is going to be crucial – and we hope to present at future meetings – is the biomarker analysis to show which are the patients who had the minor responses and had some activity," he added, with initial focus on biomarkers in the PI3 kinase/Akt/mTOR pathway.

The findings came on the heels of a phase II trial that suggested everolimus had promising efficacy in this hard-to-treat cancer (J. Clin. Oncol. 2010;28:1904-10). The PI3 kinase/Akt/mTOR pathway is a logical target in this disease, as it is dysregulated in 50%-60% of cases, according to Dr. Van Cutsem, an oncologist at University Hospital Gasthuisberg in Leuven, Belgium.

"The progression-free survival curve is somewhat provocative," said discussant Dr. David H. Ilson of the Memorial Sloan-Kettering Cancer Center in New York.

"We clearly need the study of biomarkers to identify the likely small subset of patients who may benefit from treatment with this agent," he said. "However, I would give a caveat here: This does not mean that we should mandate a biomarker up front in all of our new studies because we may miss detection of activity in all patients by mandating biomarkers up front."

Dr. Ilson noted that the magnitude of benefit may be important to taking everolimus forward in gastric cancer. "How high should the bar be set to consider developing new agents? Is a 1- to 2-month improvement in progression-free survival or overall survival enough of an adequate benchmark?" he questioned.

Patients were eligible for GRANITE-1 if they had advanced gastric cancer (or gastroesophageal junction cancer, provided the majority of the tumor was in the stomach) and had experienced progression after one or two lines of systemic chemotherapy.

In all, 656 patients were randomized 2:1 to everolimus 10 mg daily or placebo, each along with best supportive care. They had a median age of 62 years, and about 55% were from Asia. They were almost evenly split as far as having received one or two prior lines of chemotherapy.

The median duration of treatment was 7.1 weeks with everolimus and 6.4 weeks with placebo, Dr. Van Cutsem reported.

"The safety profile of everolimus was similar to that observed with everolimus in other cancers, with no new safety signals identified," he said. The rate of any grade 3 or 4 adverse event was 71% with everolimus and 54% with placebo, with the largest differences seen for certain gastrointestinal and hematologic adverse events.

Median overall survival, the trial’s primary end point, was 5.4 months with everolimus and 4.3 months with placebo, a nonsignificant difference. Subgroup analyses failed to identify any specific subgroup that had significant benefit.

Median progression-free survival was 1.7 months with everolimus and 1.4 months with placebo (hazard ratio, 0.66; P less than .0001). The overall response rate was 4.5% and 2.1%, respectively.

Dr. Van Cutsem disclosed that he receives research funding from Novartis. Dr. Ilson disclosed that he receives research funding from Bayer, BMS-ImClone, Roche-Genentech, and Sanofi-Aventis.

SAN FRANCISCO – Despite disappointing results in an international randomized phase III trial, investigators are hoping a biomarker will help them to find a role for everolimus in advanced gastric cancer.

The oral mammalian target of rapamycin (mTOR) inhibitor appeared to have limited efficacy as second- or third-line therapy in the GRANITE-1 study of more than 650 patients.

Overall survival was statistically indistinguishable between patients given everolimus (Afinitor) and those given a placebo, each added to best supportive care. Statistically, progression-free survival was significantly better with everolimus, but the absolute benefit was small, just 0.3 months.

"This trial was negative," Dr. Eric Van Cutsem told attendees when he reported the results at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology. But the story did not end there.

"What is going to be crucial – and we hope to present at future meetings – is the biomarker analysis to show which are the patients who had the minor responses and had some activity," he added, with initial focus on biomarkers in the PI3 kinase/Akt/mTOR pathway.

The findings came on the heels of a phase II trial that suggested everolimus had promising efficacy in this hard-to-treat cancer (J. Clin. Oncol. 2010;28:1904-10). The PI3 kinase/Akt/mTOR pathway is a logical target in this disease, as it is dysregulated in 50%-60% of cases, according to Dr. Van Cutsem, an oncologist at University Hospital Gasthuisberg in Leuven, Belgium.

"The progression-free survival curve is somewhat provocative," said discussant Dr. David H. Ilson of the Memorial Sloan-Kettering Cancer Center in New York.

"We clearly need the study of biomarkers to identify the likely small subset of patients who may benefit from treatment with this agent," he said. "However, I would give a caveat here: This does not mean that we should mandate a biomarker up front in all of our new studies because we may miss detection of activity in all patients by mandating biomarkers up front."

Dr. Ilson noted that the magnitude of benefit may be important to taking everolimus forward in gastric cancer. "How high should the bar be set to consider developing new agents? Is a 1- to 2-month improvement in progression-free survival or overall survival enough of an adequate benchmark?" he questioned.

Patients were eligible for GRANITE-1 if they had advanced gastric cancer (or gastroesophageal junction cancer, provided the majority of the tumor was in the stomach) and had experienced progression after one or two lines of systemic chemotherapy.

In all, 656 patients were randomized 2:1 to everolimus 10 mg daily or placebo, each along with best supportive care. They had a median age of 62 years, and about 55% were from Asia. They were almost evenly split as far as having received one or two prior lines of chemotherapy.

The median duration of treatment was 7.1 weeks with everolimus and 6.4 weeks with placebo, Dr. Van Cutsem reported.

"The safety profile of everolimus was similar to that observed with everolimus in other cancers, with no new safety signals identified," he said. The rate of any grade 3 or 4 adverse event was 71% with everolimus and 54% with placebo, with the largest differences seen for certain gastrointestinal and hematologic adverse events.

Median overall survival, the trial’s primary end point, was 5.4 months with everolimus and 4.3 months with placebo, a nonsignificant difference. Subgroup analyses failed to identify any specific subgroup that had significant benefit.

Median progression-free survival was 1.7 months with everolimus and 1.4 months with placebo (hazard ratio, 0.66; P less than .0001). The overall response rate was 4.5% and 2.1%, respectively.

Dr. Van Cutsem disclosed that he receives research funding from Novartis. Dr. Ilson disclosed that he receives research funding from Bayer, BMS-ImClone, Roche-Genentech, and Sanofi-Aventis.

SAN FRANCISCO – Despite disappointing results in an international randomized phase III trial, investigators are hoping a biomarker will help them to find a role for everolimus in advanced gastric cancer.

The oral mammalian target of rapamycin (mTOR) inhibitor appeared to have limited efficacy as second- or third-line therapy in the GRANITE-1 study of more than 650 patients.

Overall survival was statistically indistinguishable between patients given everolimus (Afinitor) and those given a placebo, each added to best supportive care. Statistically, progression-free survival was significantly better with everolimus, but the absolute benefit was small, just 0.3 months.

"This trial was negative," Dr. Eric Van Cutsem told attendees when he reported the results at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology. But the story did not end there.

"What is going to be crucial – and we hope to present at future meetings – is the biomarker analysis to show which are the patients who had the minor responses and had some activity," he added, with initial focus on biomarkers in the PI3 kinase/Akt/mTOR pathway.

The findings came on the heels of a phase II trial that suggested everolimus had promising efficacy in this hard-to-treat cancer (J. Clin. Oncol. 2010;28:1904-10). The PI3 kinase/Akt/mTOR pathway is a logical target in this disease, as it is dysregulated in 50%-60% of cases, according to Dr. Van Cutsem, an oncologist at University Hospital Gasthuisberg in Leuven, Belgium.

"The progression-free survival curve is somewhat provocative," said discussant Dr. David H. Ilson of the Memorial Sloan-Kettering Cancer Center in New York.

"We clearly need the study of biomarkers to identify the likely small subset of patients who may benefit from treatment with this agent," he said. "However, I would give a caveat here: This does not mean that we should mandate a biomarker up front in all of our new studies because we may miss detection of activity in all patients by mandating biomarkers up front."

Dr. Ilson noted that the magnitude of benefit may be important to taking everolimus forward in gastric cancer. "How high should the bar be set to consider developing new agents? Is a 1- to 2-month improvement in progression-free survival or overall survival enough of an adequate benchmark?" he questioned.

Patients were eligible for GRANITE-1 if they had advanced gastric cancer (or gastroesophageal junction cancer, provided the majority of the tumor was in the stomach) and had experienced progression after one or two lines of systemic chemotherapy.

In all, 656 patients were randomized 2:1 to everolimus 10 mg daily or placebo, each along with best supportive care. They had a median age of 62 years, and about 55% were from Asia. They were almost evenly split as far as having received one or two prior lines of chemotherapy.

The median duration of treatment was 7.1 weeks with everolimus and 6.4 weeks with placebo, Dr. Van Cutsem reported.

"The safety profile of everolimus was similar to that observed with everolimus in other cancers, with no new safety signals identified," he said. The rate of any grade 3 or 4 adverse event was 71% with everolimus and 54% with placebo, with the largest differences seen for certain gastrointestinal and hematologic adverse events.

Median overall survival, the trial’s primary end point, was 5.4 months with everolimus and 4.3 months with placebo, a nonsignificant difference. Subgroup analyses failed to identify any specific subgroup that had significant benefit.

Median progression-free survival was 1.7 months with everolimus and 1.4 months with placebo (hazard ratio, 0.66; P less than .0001). The overall response rate was 4.5% and 2.1%, respectively.

Dr. Van Cutsem disclosed that he receives research funding from Novartis. Dr. Ilson disclosed that he receives research funding from Bayer, BMS-ImClone, Roche-Genentech, and Sanofi-Aventis.

SAN FRANCISCO – The safety and efficacy profile of aspirin for preventing esophageal and other cancers is likely to limit its use to people at highest risk, said Dr. Janusz A. Jankowski in a presentation at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology.

"The commonest question I’m asked is, ‘Should I now be taking low-dose aspirin?’," he said. "And my answer to that at this moment in time is indefatigably, no, you should not, unless you have a secondary risk factor," such as Barrett metaplasia or colon polyps (or a secondary cardiovascular risk factor).

Collectively, evidence suggests that taking aspirin reduces cancer risk by about one quarter, according to Dr. Jankowski; however, aspirin has to be taken for at least a decade to see clinically meaningful benefit (Lancet 2011;377:31-41). "Chemoprevention is a very, very long term goal here," he commented. "This is not about taking it for a few years and hoping to get a benefit – it’s 10 to 20 years. And that means that the side effect profile ... must be exceptionally low."

The risk of bleeding with aspirin is well known. Rates of any upper gastrointestinal bleeding and serious GI bleeding on the medication are about 4% and 2% per year, respectively, and aspirin particularly increases the risk of spontaneous GI bleeding after age 50 years. However, adding proton pump inhibitor (PPI) therapy largely attenuates the excess risk.

But new data have also implicated aspirin in the development of macular degeneration (Ophthalmology 2012;119:112-8). "That’s another thing that you may have to consider ... Could aspirin be causing something else, as well as preventing something?" said Dr. Jankowski, a visiting professor of GI oncology at the University of Oxford (U.K.) and the Honorary Sir James Black Professor at Barts and the London School of Medicine and Dentistry.

The authors of a recent analysis concluded that healthy people do not experience any significant reduction in either cardiovascular or cancer deaths when they take aspirin (Arch. Intern. Med. 2012;172:209-16), which "has kind of dashed our hopes for primary prevention," he said.

In the secondary prevention realm, the randomized Aspect trial (a phase III, randomized, study of aspirin and esomeprazole chemoprevention in Barrett’s metaplasia) is comparing lower- vs. higher-dose PPI therapy using esomeprazole. Each dosage level of esomeprazole is given with or without aspirin in about 2,500 patients with Barrett’s esophagus, which is thought to carry a 3%-5% lifetime risk of esophageal cancer.

In this trial, results for cancer and other clinical outcomes are not expected until 2019. But early endoscopic findings suggest that the added aspirin helps promote the appearance of new squamous epithelial islands in areas previously showing Barrett metaplasia, a promising surrogate end point, according to Dr. Jankowski.

At the same time, he cautioned against overreliance on aspirin, saying, "As good as aspirin is, ... it’s still not good enough because in fact we still don’t know who’s going to get a response, and we still think that the best response rate is probably only 20% in the population."

BADCAT (the Barrett’s Dysplasia and Cancer Task Force), which exhaustively reviews evidence on the topic, has identified aspirin effectiveness and responsiveness as an area of high priority for research. And because genetics have a major role here, the Chopin study (chemoprevention of premalignant intestinal neoplasia) is assessing genomewide associations in the development and prevention of esophageal cancer, including identification of aspirin-response genes.

Summing up the evidence on the use of aspirin to prevent esophageal and other cancers, Dr. Jankowski predicted, "In the next 2 years, there may very well be a recommendation that if you have secondary risk factors for cancer, like Barrett’s ... that you may be given aspirin therapy in addition to your proton pump inhibitor."

Only people in a fairly narrow age range will have net benefit, based on currently available data. "Aspirin is a wonderful agent. Having said that, there is no evidence whatsoever that even if you’ve got secondary risks factors for cardiac disease or cancer, you should take your aspirin before 55 years of age," he explained. And starting aspirin after age 75 years is not justified given the current human lifespan and the need for prolonged use to achieve benefit.

"The window of taking aspirin is between 55 and 75," he concluded, adding "while I am an enthusiast of aspirin ... it is only part of the jigsaw, and we still need even to answer that part."

Dr. Jankowski disclosed that he is a consultant to and receives research funding and honoraria from AstraZeneca.

SAN FRANCISCO – The safety and efficacy profile of aspirin for preventing esophageal and other cancers is likely to limit its use to people at highest risk, said Dr. Janusz A. Jankowski in a presentation at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology.

"The commonest question I’m asked is, ‘Should I now be taking low-dose aspirin?’," he said. "And my answer to that at this moment in time is indefatigably, no, you should not, unless you have a secondary risk factor," such as Barrett metaplasia or colon polyps (or a secondary cardiovascular risk factor).

Collectively, evidence suggests that taking aspirin reduces cancer risk by about one quarter, according to Dr. Jankowski; however, aspirin has to be taken for at least a decade to see clinically meaningful benefit (Lancet 2011;377:31-41). "Chemoprevention is a very, very long term goal here," he commented. "This is not about taking it for a few years and hoping to get a benefit – it’s 10 to 20 years. And that means that the side effect profile ... must be exceptionally low."

The risk of bleeding with aspirin is well known. Rates of any upper gastrointestinal bleeding and serious GI bleeding on the medication are about 4% and 2% per year, respectively, and aspirin particularly increases the risk of spontaneous GI bleeding after age 50 years. However, adding proton pump inhibitor (PPI) therapy largely attenuates the excess risk.

But new data have also implicated aspirin in the development of macular degeneration (Ophthalmology 2012;119:112-8). "That’s another thing that you may have to consider ... Could aspirin be causing something else, as well as preventing something?" said Dr. Jankowski, a visiting professor of GI oncology at the University of Oxford (U.K.) and the Honorary Sir James Black Professor at Barts and the London School of Medicine and Dentistry.

The authors of a recent analysis concluded that healthy people do not experience any significant reduction in either cardiovascular or cancer deaths when they take aspirin (Arch. Intern. Med. 2012;172:209-16), which "has kind of dashed our hopes for primary prevention," he said.

In the secondary prevention realm, the randomized Aspect trial (a phase III, randomized, study of aspirin and esomeprazole chemoprevention in Barrett’s metaplasia) is comparing lower- vs. higher-dose PPI therapy using esomeprazole. Each dosage level of esomeprazole is given with or without aspirin in about 2,500 patients with Barrett’s esophagus, which is thought to carry a 3%-5% lifetime risk of esophageal cancer.

In this trial, results for cancer and other clinical outcomes are not expected until 2019. But early endoscopic findings suggest that the added aspirin helps promote the appearance of new squamous epithelial islands in areas previously showing Barrett metaplasia, a promising surrogate end point, according to Dr. Jankowski.

At the same time, he cautioned against overreliance on aspirin, saying, "As good as aspirin is, ... it’s still not good enough because in fact we still don’t know who’s going to get a response, and we still think that the best response rate is probably only 20% in the population."

BADCAT (the Barrett’s Dysplasia and Cancer Task Force), which exhaustively reviews evidence on the topic, has identified aspirin effectiveness and responsiveness as an area of high priority for research. And because genetics have a major role here, the Chopin study (chemoprevention of premalignant intestinal neoplasia) is assessing genomewide associations in the development and prevention of esophageal cancer, including identification of aspirin-response genes.

Summing up the evidence on the use of aspirin to prevent esophageal and other cancers, Dr. Jankowski predicted, "In the next 2 years, there may very well be a recommendation that if you have secondary risk factors for cancer, like Barrett’s ... that you may be given aspirin therapy in addition to your proton pump inhibitor."

Only people in a fairly narrow age range will have net benefit, based on currently available data. "Aspirin is a wonderful agent. Having said that, there is no evidence whatsoever that even if you’ve got secondary risks factors for cardiac disease or cancer, you should take your aspirin before 55 years of age," he explained. And starting aspirin after age 75 years is not justified given the current human lifespan and the need for prolonged use to achieve benefit.

"The window of taking aspirin is between 55 and 75," he concluded, adding "while I am an enthusiast of aspirin ... it is only part of the jigsaw, and we still need even to answer that part."

Dr. Jankowski disclosed that he is a consultant to and receives research funding and honoraria from AstraZeneca.

SAN FRANCISCO – The safety and efficacy profile of aspirin for preventing esophageal and other cancers is likely to limit its use to people at highest risk, said Dr. Janusz A. Jankowski in a presentation at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology.

"The commonest question I’m asked is, ‘Should I now be taking low-dose aspirin?’," he said. "And my answer to that at this moment in time is indefatigably, no, you should not, unless you have a secondary risk factor," such as Barrett metaplasia or colon polyps (or a secondary cardiovascular risk factor).

Collectively, evidence suggests that taking aspirin reduces cancer risk by about one quarter, according to Dr. Jankowski; however, aspirin has to be taken for at least a decade to see clinically meaningful benefit (Lancet 2011;377:31-41). "Chemoprevention is a very, very long term goal here," he commented. "This is not about taking it for a few years and hoping to get a benefit – it’s 10 to 20 years. And that means that the side effect profile ... must be exceptionally low."

The risk of bleeding with aspirin is well known. Rates of any upper gastrointestinal bleeding and serious GI bleeding on the medication are about 4% and 2% per year, respectively, and aspirin particularly increases the risk of spontaneous GI bleeding after age 50 years. However, adding proton pump inhibitor (PPI) therapy largely attenuates the excess risk.

But new data have also implicated aspirin in the development of macular degeneration (Ophthalmology 2012;119:112-8). "That’s another thing that you may have to consider ... Could aspirin be causing something else, as well as preventing something?" said Dr. Jankowski, a visiting professor of GI oncology at the University of Oxford (U.K.) and the Honorary Sir James Black Professor at Barts and the London School of Medicine and Dentistry.

The authors of a recent analysis concluded that healthy people do not experience any significant reduction in either cardiovascular or cancer deaths when they take aspirin (Arch. Intern. Med. 2012;172:209-16), which "has kind of dashed our hopes for primary prevention," he said.

In the secondary prevention realm, the randomized Aspect trial (a phase III, randomized, study of aspirin and esomeprazole chemoprevention in Barrett’s metaplasia) is comparing lower- vs. higher-dose PPI therapy using esomeprazole. Each dosage level of esomeprazole is given with or without aspirin in about 2,500 patients with Barrett’s esophagus, which is thought to carry a 3%-5% lifetime risk of esophageal cancer.

In this trial, results for cancer and other clinical outcomes are not expected until 2019. But early endoscopic findings suggest that the added aspirin helps promote the appearance of new squamous epithelial islands in areas previously showing Barrett metaplasia, a promising surrogate end point, according to Dr. Jankowski.

At the same time, he cautioned against overreliance on aspirin, saying, "As good as aspirin is, ... it’s still not good enough because in fact we still don’t know who’s going to get a response, and we still think that the best response rate is probably only 20% in the population."

BADCAT (the Barrett’s Dysplasia and Cancer Task Force), which exhaustively reviews evidence on the topic, has identified aspirin effectiveness and responsiveness as an area of high priority for research. And because genetics have a major role here, the Chopin study (chemoprevention of premalignant intestinal neoplasia) is assessing genomewide associations in the development and prevention of esophageal cancer, including identification of aspirin-response genes.

Summing up the evidence on the use of aspirin to prevent esophageal and other cancers, Dr. Jankowski predicted, "In the next 2 years, there may very well be a recommendation that if you have secondary risk factors for cancer, like Barrett’s ... that you may be given aspirin therapy in addition to your proton pump inhibitor."

Only people in a fairly narrow age range will have net benefit, based on currently available data. "Aspirin is a wonderful agent. Having said that, there is no evidence whatsoever that even if you’ve got secondary risks factors for cardiac disease or cancer, you should take your aspirin before 55 years of age," he explained. And starting aspirin after age 75 years is not justified given the current human lifespan and the need for prolonged use to achieve benefit.

"The window of taking aspirin is between 55 and 75," he concluded, adding "while I am an enthusiast of aspirin ... it is only part of the jigsaw, and we still need even to answer that part."

Dr. Jankowski disclosed that he is a consultant to and receives research funding and honoraria from AstraZeneca.

SAN FRANCISCO – Patients with gastrointestinal stromal tumors have a poorer prognosis if their primary tumor is located outside the gastrointestinal tract, researchers reported.

In an analysis of a U.S. population–based cohort of 2,591 patients who underwent surgery for gastrointestinal stromal tumors (GIST), 10% had primaries in the omentum, mesentery, retroperitoneum, or another location separate from the tubular gastrointestinal tract.

The 5-year overall survival rate for this group of patients with extragastrointestinal stromal tumors (EGIST) was 62%, which was significantly poorer than the 70% observed for those with primaries in the typical gastrointestinal tract location, mainly the stomach or small intestine.

Patients with primaries outside the gastrointestinal tract had larger tumors and more advanced stage, the study found. But even after adjustment for these potential confounders, extragastrointestinal location was still associated with a nearly 30% increase in the risk of death.

"The results of our study suggest that tumor location outside the gastrointestinal tract should be considered as an important variable, and risk stratification for prognosis should account for these rare GIST locations," said lead investigator Dr. Mary L. Guye, who presented the data at the gastrointestinal cancers symposium sponsored by the American Society of Clinical Oncology.

The reason for this poorer prognosis is unknown, said Dr. Guye, a surgical oncologist at the City of Hope National Medical Center in Duarte, Calif.

"It could be these are truly metastatic foci with an unidentified or unknown primary tumor, or they may simply be more aggressive tumors that arise de novo outside the gastrointestinal tract," she proposed, calling for further studies to investigate the tumor biology of these rare GIST tumors.

"Evaluation of histology, looking at tumor mitotic rate, and also mutational analyses may help to determine why extra-intestinal location confers a worse prognosis for GIST tumors," she said.

In an interview, Dr. Allan Pickens of Emory University in Atlanta, cochair of the scientific session in which the data were presented, said, "This was a great study [that further defined] the difference between the traditional GISTs that we treat more commonly and the ones that are outside of the gastrointestinal tract [and] have a worse prognosis."

Dr. Pickens agreed that molecular analyses will help better define how the two groups of tumors differ. "It would be interesting to see some of the microRNA information applied to those different tumors to see if it could predict those worse outcomes," he said.

Dr. Guye and her colleagues analyzed SEER (Surveillance, Epidemiology, and End Results) data for 1996-2008 to identify their study population with GIST, choosing the start year because that is when diagnosis of this tumor became more accurate, she explained.

Patients with primaries in and outside the gastrointestinal tract had similar age and racial distribution. But patients with EGIST were more likely to receive radiation therapy, to have tumors larger than 10 cm, and to have regional or distant disease.

Median overall survival was 120 months in the group with primaries in the gastrointestinal tract, and 105 months in the group with primaries in other locations (P = .002).

Additional analyses showed that location outside the gastrointestinal tract conferred a worse prognosis, even when tumors were matched for size and stage of disease, Dr. Guye reported. And the association stood up in multivariate analysis (P = .03).

However, "the SEER database does not include information on mitotic rate or information on systemic medical therapies," she noted. "Therefore, we were not able to include these factors in our analysis."

Dr. Guye and Dr. Pickens disclosed that they had no relevant conflicts of interest.

SAN FRANCISCO – Patients with gastrointestinal stromal tumors have a poorer prognosis if their primary tumor is located outside the gastrointestinal tract, researchers reported.

In an analysis of a U.S. population–based cohort of 2,591 patients who underwent surgery for gastrointestinal stromal tumors (GIST), 10% had primaries in the omentum, mesentery, retroperitoneum, or another location separate from the tubular gastrointestinal tract.

The 5-year overall survival rate for this group of patients with extragastrointestinal stromal tumors (EGIST) was 62%, which was significantly poorer than the 70% observed for those with primaries in the typical gastrointestinal tract location, mainly the stomach or small intestine.

Patients with primaries outside the gastrointestinal tract had larger tumors and more advanced stage, the study found. But even after adjustment for these potential confounders, extragastrointestinal location was still associated with a nearly 30% increase in the risk of death.

"The results of our study suggest that tumor location outside the gastrointestinal tract should be considered as an important variable, and risk stratification for prognosis should account for these rare GIST locations," said lead investigator Dr. Mary L. Guye, who presented the data at the gastrointestinal cancers symposium sponsored by the American Society of Clinical Oncology.

The reason for this poorer prognosis is unknown, said Dr. Guye, a surgical oncologist at the City of Hope National Medical Center in Duarte, Calif.

"It could be these are truly metastatic foci with an unidentified or unknown primary tumor, or they may simply be more aggressive tumors that arise de novo outside the gastrointestinal tract," she proposed, calling for further studies to investigate the tumor biology of these rare GIST tumors.

"Evaluation of histology, looking at tumor mitotic rate, and also mutational analyses may help to determine why extra-intestinal location confers a worse prognosis for GIST tumors," she said.

In an interview, Dr. Allan Pickens of Emory University in Atlanta, cochair of the scientific session in which the data were presented, said, "This was a great study [that further defined] the difference between the traditional GISTs that we treat more commonly and the ones that are outside of the gastrointestinal tract [and] have a worse prognosis."

Dr. Pickens agreed that molecular analyses will help better define how the two groups of tumors differ. "It would be interesting to see some of the microRNA information applied to those different tumors to see if it could predict those worse outcomes," he said.

Dr. Guye and her colleagues analyzed SEER (Surveillance, Epidemiology, and End Results) data for 1996-2008 to identify their study population with GIST, choosing the start year because that is when diagnosis of this tumor became more accurate, she explained.

Patients with primaries in and outside the gastrointestinal tract had similar age and racial distribution. But patients with EGIST were more likely to receive radiation therapy, to have tumors larger than 10 cm, and to have regional or distant disease.

Median overall survival was 120 months in the group with primaries in the gastrointestinal tract, and 105 months in the group with primaries in other locations (P = .002).

Additional analyses showed that location outside the gastrointestinal tract conferred a worse prognosis, even when tumors were matched for size and stage of disease, Dr. Guye reported. And the association stood up in multivariate analysis (P = .03).

However, "the SEER database does not include information on mitotic rate or information on systemic medical therapies," she noted. "Therefore, we were not able to include these factors in our analysis."

Dr. Guye and Dr. Pickens disclosed that they had no relevant conflicts of interest.

SAN FRANCISCO – Patients with gastrointestinal stromal tumors have a poorer prognosis if their primary tumor is located outside the gastrointestinal tract, researchers reported.

In an analysis of a U.S. population–based cohort of 2,591 patients who underwent surgery for gastrointestinal stromal tumors (GIST), 10% had primaries in the omentum, mesentery, retroperitoneum, or another location separate from the tubular gastrointestinal tract.

The 5-year overall survival rate for this group of patients with extragastrointestinal stromal tumors (EGIST) was 62%, which was significantly poorer than the 70% observed for those with primaries in the typical gastrointestinal tract location, mainly the stomach or small intestine.

Patients with primaries outside the gastrointestinal tract had larger tumors and more advanced stage, the study found. But even after adjustment for these potential confounders, extragastrointestinal location was still associated with a nearly 30% increase in the risk of death.

"The results of our study suggest that tumor location outside the gastrointestinal tract should be considered as an important variable, and risk stratification for prognosis should account for these rare GIST locations," said lead investigator Dr. Mary L. Guye, who presented the data at the gastrointestinal cancers symposium sponsored by the American Society of Clinical Oncology.

The reason for this poorer prognosis is unknown, said Dr. Guye, a surgical oncologist at the City of Hope National Medical Center in Duarte, Calif.

"It could be these are truly metastatic foci with an unidentified or unknown primary tumor, or they may simply be more aggressive tumors that arise de novo outside the gastrointestinal tract," she proposed, calling for further studies to investigate the tumor biology of these rare GIST tumors.

"Evaluation of histology, looking at tumor mitotic rate, and also mutational analyses may help to determine why extra-intestinal location confers a worse prognosis for GIST tumors," she said.

In an interview, Dr. Allan Pickens of Emory University in Atlanta, cochair of the scientific session in which the data were presented, said, "This was a great study [that further defined] the difference between the traditional GISTs that we treat more commonly and the ones that are outside of the gastrointestinal tract [and] have a worse prognosis."

Dr. Pickens agreed that molecular analyses will help better define how the two groups of tumors differ. "It would be interesting to see some of the microRNA information applied to those different tumors to see if it could predict those worse outcomes," he said.

Dr. Guye and her colleagues analyzed SEER (Surveillance, Epidemiology, and End Results) data for 1996-2008 to identify their study population with GIST, choosing the start year because that is when diagnosis of this tumor became more accurate, she explained.

Patients with primaries in and outside the gastrointestinal tract had similar age and racial distribution. But patients with EGIST were more likely to receive radiation therapy, to have tumors larger than 10 cm, and to have regional or distant disease.

Median overall survival was 120 months in the group with primaries in the gastrointestinal tract, and 105 months in the group with primaries in other locations (P = .002).

Additional analyses showed that location outside the gastrointestinal tract conferred a worse prognosis, even when tumors were matched for size and stage of disease, Dr. Guye reported. And the association stood up in multivariate analysis (P = .03).

However, "the SEER database does not include information on mitotic rate or information on systemic medical therapies," she noted. "Therefore, we were not able to include these factors in our analysis."

Dr. Guye and Dr. Pickens disclosed that they had no relevant conflicts of interest.

SAN FRANCISCO – A small proportion of patients with a lower-risk intraductal papillary mucinous neoplasm go on to develop pancreatic ductal adenocarcinoma in an entirely separate region of the gland, an outcome that raises questions about the surveillance strategy for these patients.

Dr. Jennifer LaFemina and her colleagues at the Memorial Sloan-Kettering Cancer Center, New York, retrospectively examined data for 158 patients who were seen at the center in 1989-2010 and who had intraductal papillary mucinous neoplasms (IPMNs) with favorable enough initial features that they were followed with radiographic surveillance for at least 6 months after diagnosis, instead of being taken right to surgery.

Five patients (3.2%) developed pancreatic cancer in a different part of the gland, she reported at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology. In all cases, the cancer was stage 2A or 2B at diagnosis and was resected with negative margins; however, with a median follow-up of just 1 year, three of the five patients have died.

Dr. LaFemina noted that several published studies have shown that IPMN represents a field defect of ductal instability, and although many have focused on the risk of developing malignancy in the IPMN cyst itself, "the risk of developing a cancer in a region of the pancreas separate from the target cyst that we are monitoring is poorly defined."

In her study, 97 of the 158 patients eventually underwent resection at a median of 16 months after IPMN diagnosis because of a worrisome change either in the IPMN target cyst or elsewhere in the pancreas (J. Clin. Oncol. 2012;30 [suppl 4]:abstract 152). "Five patients ultimately came to surgery not because of the target cyst but because of a ductal adenocarcinoma without IPMN that developed while we were watching them," Dr. LaFemina explained. (In all, 13 of the patients having resection were found to have cancer at the IPMN target cyst.)

In four of these five cases of a separate cancer, the IPMN that had been under surveillance was of the branched-duct subtype and was smaller, measuring no more than 1.6 cm. And in three cases, the time from diagnosis of the IPMN to diagnosis of the separate cancer was fairly short, just 16 months or less.

"Diagnostic, operative, and surveillance strategies should consider the cancer risk not only to the [IPMN] target cyst that we are presented with, but also to the entire gland," Dr. LaFemina recommended. Although the center has not changed its surveillance strategy for these patients, she said that this new information raises some concerns.

The findings also bring up the question of whether to recommend total pancreatectomy for patients who undergo resection, to be on the safe side. "At this point, we are not endorsing total pancreatectomy for a number of reasons," she said – notably because of the difficulty of predicting preoperatively what is actually going on in the gland.

A key question is to determine what is different about patients who develop an adenocarcinoma separate from the IPMN. "Can we identify them and watch them [more closely]? We don’t have that answer yet," Dr. LaFemina noted.

The study "has really thrown us a curve ball in terms of how we deal with these patients," commented session cochair Dr. Mitchell Posner, a surgical oncologist at the University of Chicago Medical Center.

"It’s scary, because the criteria we are using to decide not to operate on patients – which I think are appropriate – unfortunately aren’t definitively ruling out patients who will develop a pancreatic cancer in a relatively short period of time," he said in an interview. Furthermore, he noted, it’s tricky to decide whether to operate (because of the morbidity and mortality risks associated with a pancreatic resection) or to risk letting a patient develop a pancreatic cancer, knowing that an intervention potentially could have changed the outcome.

"So it is very frightening that we unfortunately don’t have the tools to identify patients accurately at this point. And we are detecting these lesions [IPMNs] at a very high frequency, so it’s a problem," Dr. Posner said. The lesions’ molecular signatures likely hold the key to identifying the bad actors, which may help target pancreatectomy to the patients who need it, he proposed.

Dr. LaFemina and Dr. Posner disclosed that they had no relevant conflicts of interest.

SAN FRANCISCO – A small proportion of patients with a lower-risk intraductal papillary mucinous neoplasm go on to develop pancreatic ductal adenocarcinoma in an entirely separate region of the gland, an outcome that raises questions about the surveillance strategy for these patients.

Dr. Jennifer LaFemina and her colleagues at the Memorial Sloan-Kettering Cancer Center, New York, retrospectively examined data for 158 patients who were seen at the center in 1989-2010 and who had intraductal papillary mucinous neoplasms (IPMNs) with favorable enough initial features that they were followed with radiographic surveillance for at least 6 months after diagnosis, instead of being taken right to surgery.

Five patients (3.2%) developed pancreatic cancer in a different part of the gland, she reported at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology. In all cases, the cancer was stage 2A or 2B at diagnosis and was resected with negative margins; however, with a median follow-up of just 1 year, three of the five patients have died.

Dr. LaFemina noted that several published studies have shown that IPMN represents a field defect of ductal instability, and although many have focused on the risk of developing malignancy in the IPMN cyst itself, "the risk of developing a cancer in a region of the pancreas separate from the target cyst that we are monitoring is poorly defined."

In her study, 97 of the 158 patients eventually underwent resection at a median of 16 months after IPMN diagnosis because of a worrisome change either in the IPMN target cyst or elsewhere in the pancreas (J. Clin. Oncol. 2012;30 [suppl 4]:abstract 152). "Five patients ultimately came to surgery not because of the target cyst but because of a ductal adenocarcinoma without IPMN that developed while we were watching them," Dr. LaFemina explained. (In all, 13 of the patients having resection were found to have cancer at the IPMN target cyst.)

In four of these five cases of a separate cancer, the IPMN that had been under surveillance was of the branched-duct subtype and was smaller, measuring no more than 1.6 cm. And in three cases, the time from diagnosis of the IPMN to diagnosis of the separate cancer was fairly short, just 16 months or less.

"Diagnostic, operative, and surveillance strategies should consider the cancer risk not only to the [IPMN] target cyst that we are presented with, but also to the entire gland," Dr. LaFemina recommended. Although the center has not changed its surveillance strategy for these patients, she said that this new information raises some concerns.

The findings also bring up the question of whether to recommend total pancreatectomy for patients who undergo resection, to be on the safe side. "At this point, we are not endorsing total pancreatectomy for a number of reasons," she said – notably because of the difficulty of predicting preoperatively what is actually going on in the gland.

A key question is to determine what is different about patients who develop an adenocarcinoma separate from the IPMN. "Can we identify them and watch them [more closely]? We don’t have that answer yet," Dr. LaFemina noted.

The study "has really thrown us a curve ball in terms of how we deal with these patients," commented session cochair Dr. Mitchell Posner, a surgical oncologist at the University of Chicago Medical Center.

"It’s scary, because the criteria we are using to decide not to operate on patients – which I think are appropriate – unfortunately aren’t definitively ruling out patients who will develop a pancreatic cancer in a relatively short period of time," he said in an interview. Furthermore, he noted, it’s tricky to decide whether to operate (because of the morbidity and mortality risks associated with a pancreatic resection) or to risk letting a patient develop a pancreatic cancer, knowing that an intervention potentially could have changed the outcome.

"So it is very frightening that we unfortunately don’t have the tools to identify patients accurately at this point. And we are detecting these lesions [IPMNs] at a very high frequency, so it’s a problem," Dr. Posner said. The lesions’ molecular signatures likely hold the key to identifying the bad actors, which may help target pancreatectomy to the patients who need it, he proposed.

Dr. LaFemina and Dr. Posner disclosed that they had no relevant conflicts of interest.

SAN FRANCISCO – A small proportion of patients with a lower-risk intraductal papillary mucinous neoplasm go on to develop pancreatic ductal adenocarcinoma in an entirely separate region of the gland, an outcome that raises questions about the surveillance strategy for these patients.

Dr. Jennifer LaFemina and her colleagues at the Memorial Sloan-Kettering Cancer Center, New York, retrospectively examined data for 158 patients who were seen at the center in 1989-2010 and who had intraductal papillary mucinous neoplasms (IPMNs) with favorable enough initial features that they were followed with radiographic surveillance for at least 6 months after diagnosis, instead of being taken right to surgery.

Five patients (3.2%) developed pancreatic cancer in a different part of the gland, she reported at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology. In all cases, the cancer was stage 2A or 2B at diagnosis and was resected with negative margins; however, with a median follow-up of just 1 year, three of the five patients have died.

Dr. LaFemina noted that several published studies have shown that IPMN represents a field defect of ductal instability, and although many have focused on the risk of developing malignancy in the IPMN cyst itself, "the risk of developing a cancer in a region of the pancreas separate from the target cyst that we are monitoring is poorly defined."

In her study, 97 of the 158 patients eventually underwent resection at a median of 16 months after IPMN diagnosis because of a worrisome change either in the IPMN target cyst or elsewhere in the pancreas (J. Clin. Oncol. 2012;30 [suppl 4]:abstract 152). "Five patients ultimately came to surgery not because of the target cyst but because of a ductal adenocarcinoma without IPMN that developed while we were watching them," Dr. LaFemina explained. (In all, 13 of the patients having resection were found to have cancer at the IPMN target cyst.)

In four of these five cases of a separate cancer, the IPMN that had been under surveillance was of the branched-duct subtype and was smaller, measuring no more than 1.6 cm. And in three cases, the time from diagnosis of the IPMN to diagnosis of the separate cancer was fairly short, just 16 months or less.

"Diagnostic, operative, and surveillance strategies should consider the cancer risk not only to the [IPMN] target cyst that we are presented with, but also to the entire gland," Dr. LaFemina recommended. Although the center has not changed its surveillance strategy for these patients, she said that this new information raises some concerns.

The findings also bring up the question of whether to recommend total pancreatectomy for patients who undergo resection, to be on the safe side. "At this point, we are not endorsing total pancreatectomy for a number of reasons," she said – notably because of the difficulty of predicting preoperatively what is actually going on in the gland.

A key question is to determine what is different about patients who develop an adenocarcinoma separate from the IPMN. "Can we identify them and watch them [more closely]? We don’t have that answer yet," Dr. LaFemina noted.

The study "has really thrown us a curve ball in terms of how we deal with these patients," commented session cochair Dr. Mitchell Posner, a surgical oncologist at the University of Chicago Medical Center.

"It’s scary, because the criteria we are using to decide not to operate on patients – which I think are appropriate – unfortunately aren’t definitively ruling out patients who will develop a pancreatic cancer in a relatively short period of time," he said in an interview. Furthermore, he noted, it’s tricky to decide whether to operate (because of the morbidity and mortality risks associated with a pancreatic resection) or to risk letting a patient develop a pancreatic cancer, knowing that an intervention potentially could have changed the outcome.

"So it is very frightening that we unfortunately don’t have the tools to identify patients accurately at this point. And we are detecting these lesions [IPMNs] at a very high frequency, so it’s a problem," Dr. Posner said. The lesions’ molecular signatures likely hold the key to identifying the bad actors, which may help target pancreatectomy to the patients who need it, he proposed.

Dr. LaFemina and Dr. Posner disclosed that they had no relevant conflicts of interest.

SAN FRANCISCO – A new 86-gene signature may add to metabolic imaging in guiding early treatment decisions for esophagogastric junction cancer, researchers reported.

The leptin gene, among others, stood out as having the potential to become a clinically useful biomarker in analyses that were presented at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology.

Dr. Russell Petty of the University of Aberdeen (Scotland) and colleagues studied 182 patients with locally advanced or metastatic disease. The investigators performed FDG-PET (¹⁸fluorodeoxyglucose–positron emission tomography) imaging and CT imaging, as well as gene expression profiling, and immunohistochemistry of pretreatment tumor tissue.

Results showed that among a subgroup of 14 patients who had a PET metabolic response to the first cycle of neoadjuvant chemotherapy, a set of 86 genes in pretreatment tumor tissue distinguished those who had a radiologic response on CT after completing all cycles of that chemotherapy, compared with those who did not respond. In a variety of predictive models, this gene signature correctly predicted radiologic response in all cases.

"We have shown that gene expression profiling can subclassify FDG-PET metabolic responders into those patients that will and will not subsequently go on to have a [radiologic] response. We have also shown that combining molecular biomarkers and FDG-PET allows an optimization of response prediction," Dr. Petty said.

This is key, as PET alone has a poor positive predictive value in this setting, with only half of PET responders to neoadjuvant chemotherapy eventually showing a histopathologic response.

"While I suppose there is little doubt that if we used PET in this way to make treatment decisions, we would have a system that would be better than unstratified empirical treatment, what it still means is that quite significant proportions of our patients will go on to receive ineffective and intensely toxic therapy," he commented.

A subsequent gene enrichment and pathway analysis identified a half-dozen pathways that provided potential mechanistic explanations for the different tumor behavior. "Within the pathways identified, the adipocytokine-signaling pathway immediately caught our attention because of the strong known [epidemiologic] link between obesity and body weight, and esophagogastric adenocarcinoma," Dr. Petty noted.

Additional analyses specifically fingered the leptin gene (among others) and showed that among PET responders, leptin expression was higher in radiologic nonresponders than in responders (P = .026).

Yet, immunohistochemistry in an independent group of 154 patients showed that strong leptin staining was also associated with better survival (P = .021), an association that held up in multivariate analysis (hazard ratio, 0.85; P = .04).

In stratified analyses, patients with leptin-positive tumors derived little survival benefit from neoadjuvant chemotherapy, whereas those with leptin-negative tumors fared considerably better if they received this therapy (indeed, about as well as the positive group).

"Leptin in particular may be a molecular biomarker that is useful to combine with PET," Dr. Petty commented. "Leptin expression is associated with chemoresistance, but at the same time is also a favorable therapy-independent biomarker, so this combination of predictive and therapy-independent prognostic impacts mean that leptin has the potential to be a clinically useful biomarker."

"I really think that this is groundbreaking. ... The real future is here" when it comes to predicting treatment response in this cancer, as suggested by this study and others," session cochair Dr. Mark Krasna of the St. Joseph Medical Center in Towson, Md., commented in an interview. "The question now is, are we ready for prime time to adopt that throughout?"

At present, more research is needed, according to Dr. Krasna, such as the ongoing validation of the 86-gene signature and trials like CALGB (Cancer and Leukemia Group B) 80803 exploring response-adaptive treatment in this cancer. "I think those are the important things we can do today. We are ready for that," he concluded.

Dr. Petty and Dr. Krasna reported that they had no relevant conflicts of interest.

SAN FRANCISCO – A new 86-gene signature may add to metabolic imaging in guiding early treatment decisions for esophagogastric junction cancer, researchers reported.

The leptin gene, among others, stood out as having the potential to become a clinically useful biomarker in analyses that were presented at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology.

Dr. Russell Petty of the University of Aberdeen (Scotland) and colleagues studied 182 patients with locally advanced or metastatic disease. The investigators performed FDG-PET (¹⁸fluorodeoxyglucose–positron emission tomography) imaging and CT imaging, as well as gene expression profiling, and immunohistochemistry of pretreatment tumor tissue.

Results showed that among a subgroup of 14 patients who had a PET metabolic response to the first cycle of neoadjuvant chemotherapy, a set of 86 genes in pretreatment tumor tissue distinguished those who had a radiologic response on CT after completing all cycles of that chemotherapy, compared with those who did not respond. In a variety of predictive models, this gene signature correctly predicted radiologic response in all cases.

"We have shown that gene expression profiling can subclassify FDG-PET metabolic responders into those patients that will and will not subsequently go on to have a [radiologic] response. We have also shown that combining molecular biomarkers and FDG-PET allows an optimization of response prediction," Dr. Petty said.

This is key, as PET alone has a poor positive predictive value in this setting, with only half of PET responders to neoadjuvant chemotherapy eventually showing a histopathologic response.

"While I suppose there is little doubt that if we used PET in this way to make treatment decisions, we would have a system that would be better than unstratified empirical treatment, what it still means is that quite significant proportions of our patients will go on to receive ineffective and intensely toxic therapy," he commented.

A subsequent gene enrichment and pathway analysis identified a half-dozen pathways that provided potential mechanistic explanations for the different tumor behavior. "Within the pathways identified, the adipocytokine-signaling pathway immediately caught our attention because of the strong known [epidemiologic] link between obesity and body weight, and esophagogastric adenocarcinoma," Dr. Petty noted.

Additional analyses specifically fingered the leptin gene (among others) and showed that among PET responders, leptin expression was higher in radiologic nonresponders than in responders (P = .026).

Yet, immunohistochemistry in an independent group of 154 patients showed that strong leptin staining was also associated with better survival (P = .021), an association that held up in multivariate analysis (hazard ratio, 0.85; P = .04).

In stratified analyses, patients with leptin-positive tumors derived little survival benefit from neoadjuvant chemotherapy, whereas those with leptin-negative tumors fared considerably better if they received this therapy (indeed, about as well as the positive group).

"Leptin in particular may be a molecular biomarker that is useful to combine with PET," Dr. Petty commented. "Leptin expression is associated with chemoresistance, but at the same time is also a favorable therapy-independent biomarker, so this combination of predictive and therapy-independent prognostic impacts mean that leptin has the potential to be a clinically useful biomarker."

"I really think that this is groundbreaking. ... The real future is here" when it comes to predicting treatment response in this cancer, as suggested by this study and others," session cochair Dr. Mark Krasna of the St. Joseph Medical Center in Towson, Md., commented in an interview. "The question now is, are we ready for prime time to adopt that throughout?"

At present, more research is needed, according to Dr. Krasna, such as the ongoing validation of the 86-gene signature and trials like CALGB (Cancer and Leukemia Group B) 80803 exploring response-adaptive treatment in this cancer. "I think those are the important things we can do today. We are ready for that," he concluded.

Dr. Petty and Dr. Krasna reported that they had no relevant conflicts of interest.

SAN FRANCISCO – A new 86-gene signature may add to metabolic imaging in guiding early treatment decisions for esophagogastric junction cancer, researchers reported.

The leptin gene, among others, stood out as having the potential to become a clinically useful biomarker in analyses that were presented at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology.

Dr. Russell Petty of the University of Aberdeen (Scotland) and colleagues studied 182 patients with locally advanced or metastatic disease. The investigators performed FDG-PET (¹⁸fluorodeoxyglucose–positron emission tomography) imaging and CT imaging, as well as gene expression profiling, and immunohistochemistry of pretreatment tumor tissue.

Results showed that among a subgroup of 14 patients who had a PET metabolic response to the first cycle of neoadjuvant chemotherapy, a set of 86 genes in pretreatment tumor tissue distinguished those who had a radiologic response on CT after completing all cycles of that chemotherapy, compared with those who did not respond. In a variety of predictive models, this gene signature correctly predicted radiologic response in all cases.

"We have shown that gene expression profiling can subclassify FDG-PET metabolic responders into those patients that will and will not subsequently go on to have a [radiologic] response. We have also shown that combining molecular biomarkers and FDG-PET allows an optimization of response prediction," Dr. Petty said.

This is key, as PET alone has a poor positive predictive value in this setting, with only half of PET responders to neoadjuvant chemotherapy eventually showing a histopathologic response.

"While I suppose there is little doubt that if we used PET in this way to make treatment decisions, we would have a system that would be better than unstratified empirical treatment, what it still means is that quite significant proportions of our patients will go on to receive ineffective and intensely toxic therapy," he commented.

A subsequent gene enrichment and pathway analysis identified a half-dozen pathways that provided potential mechanistic explanations for the different tumor behavior. "Within the pathways identified, the adipocytokine-signaling pathway immediately caught our attention because of the strong known [epidemiologic] link between obesity and body weight, and esophagogastric adenocarcinoma," Dr. Petty noted.

Additional analyses specifically fingered the leptin gene (among others) and showed that among PET responders, leptin expression was higher in radiologic nonresponders than in responders (P = .026).

Yet, immunohistochemistry in an independent group of 154 patients showed that strong leptin staining was also associated with better survival (P = .021), an association that held up in multivariate analysis (hazard ratio, 0.85; P = .04).

In stratified analyses, patients with leptin-positive tumors derived little survival benefit from neoadjuvant chemotherapy, whereas those with leptin-negative tumors fared considerably better if they received this therapy (indeed, about as well as the positive group).

"Leptin in particular may be a molecular biomarker that is useful to combine with PET," Dr. Petty commented. "Leptin expression is associated with chemoresistance, but at the same time is also a favorable therapy-independent biomarker, so this combination of predictive and therapy-independent prognostic impacts mean that leptin has the potential to be a clinically useful biomarker."

"I really think that this is groundbreaking. ... The real future is here" when it comes to predicting treatment response in this cancer, as suggested by this study and others," session cochair Dr. Mark Krasna of the St. Joseph Medical Center in Towson, Md., commented in an interview. "The question now is, are we ready for prime time to adopt that throughout?"

At present, more research is needed, according to Dr. Krasna, such as the ongoing validation of the 86-gene signature and trials like CALGB (Cancer and Leukemia Group B) 80803 exploring response-adaptive treatment in this cancer. "I think those are the important things we can do today. We are ready for that," he concluded.

Dr. Petty and Dr. Krasna reported that they had no relevant conflicts of interest.

SAN FRANCISCO – Eagerly awaited results of the CO.20 trial in metastatic chemorefractory KRAS wild-type colorectal cancer are disappointing, showing no significant gain in the primary end point of overall survival from the addition of brivanib to cetuximab.

Patients in the phase III trial who received brivanib – an investigational oral inhibitor of signaling in the vascular endothelial growth factor (VEGF) and fibroblast growth factor (FGF) pathways – lived a median of about 9 months, compared with 8 months for those given placebo, Dr. Lillian L. Siu reported at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology.

Adding brivanib did prolong progression-free survival and increase response rate, but the tradeoff was more toxicity, greater treatment discontinuation, and faster deterioration in quality of life, she said.

"There is good scientific rationale to investigate [brivanib] in combination with EGFR [epidermal growth factor receptor] inhibitors" such as cetuximab, commented Dr. Siu, a senior staff physician in the division of medical oncology and hematology at Princess Margaret Hospital, Toronto. The CO.20 trial, however, "did not meet its primary end point of overall survival," she said. Analyses assessing potential biomarkers are still ongoing.

"The median does not well reflect the overall benefit in progression-free survival by the nature of the curves, which tend to splay at the 50-percentile mark," pointed out discussant Dr. Herbert Hurwitz of the Duke Cancer Institute in Durham, N.C.

"The side effect profile for this study was unanticipatedly large," he maintained, and the increased toxicity was "clinically meaningful," leading to higher dropout and to "significant dose reductions in respect to brivanib, but also in respect to cetuximab, thereby compromising not only the value of the experimental drug, but of the standard drug as well." Thus, "the toxicity of this doublet likely confounded some of the efficacy results," he added.

The trial’s findings call into question the role in colorectal cancer of basic FGF (bFGF) as a mechanism of resistance to anti-VEGF therapy, suggested by early research. "Although this [trial] does not exclude a benefit, we need to rethink exactly what is the role, particularly given the diagnostic difficulty of assessing bFGF in both plasma and tissue," Dr. Hurwitz said.

"This data clearly will not lead to a regulatory approval" of brivanib, he concluded. "Data as monotherapy are not known, and data in trials in hepatocellular cancer are still ongoing."

To be eligible for the CO.20 trial, patients were required to have previously received a thymidylate synthase inhibitor (such as 5-fluorouracil or capecitabine) and to be intolerant of or have disease refractory to irinotecan and oxaliplatin. They could have received, at most, one prior therapy targeting VEGF but not any prior therapies targeting EGFR.

In all, 750 patients were assigned evenly to weekly cetuximab (Erbitux), an antibody to EGFR, plus either daily placebo or daily brivanib.

The patients had median age of about 64 years; nearly all (92%) had received more than three prior lines of chemotherapy in any setting (neoadjuvant, adjuvant, and/or metastatic), and a large share (41%) had received a prior therapy targeting VEGF.

Main results showed that, with a median follow-up of 19 months, there was no significant gain in median overall survival with brivanib-cetuximab compared with placebo-cetuximab in intention-to-treat analyses (8.8 vs. 8.1 months). The relative benefit was essentially the same in all patient subgroups.

Brivanib was associated with better progression-free survival (5.0 vs. 3.4 months; hazard ratio, 0.72; P less than .0001), a finding consistent across subgroups, and with a higher response rate (13.6% vs. 7.2%, P = .004).

The proportions of patients receiving at least 90% of the planned dose intensities of brivanib/placebo and cetuximab were lower in the brivanib group. "This was mainly due to dose reductions as well as omissions of these drugs," Dr. Siu noted.

The overall rate of grade 3 or worse nonhematologic toxicity was higher with brivanib than with placebo (78% vs. 53%). Fatigue, hypertension, gastrointestinal toxicity, and certain biochemical abnormalities were all more common with the drug. The rate of discontinuation as a result of adverse events was also higher with brivanib (22%) than with placebo (3%).

Finally, the median time to deterioration of quality of life was shorter with brivanib than with placebo for both global health (1.1 vs. 1.6 months, P = .02) and physical function (1.7 vs. 5.6 months, P less than .0001).

Dr. Siu reported that she receives research funding from Bristol-Myers Squibb. Dr. Hurwitz reported that he is a consultant to Bristol-Myers Squibb and Genentech/Roche; receives honoraria from Roche; and receives research funding from Acceleron Pharma, Amgen, Bristol-Myers Squibb, Cephalon, Genentech, GlaxoSmithKline, Novartis, Pfizer, Roche, Sanofi-Aventis, Threshold, and Tracon.

SAN FRANCISCO – Eagerly awaited results of the CO.20 trial in metastatic chemorefractory KRAS wild-type colorectal cancer are disappointing, showing no significant gain in the primary end point of overall survival from the addition of brivanib to cetuximab.

Patients in the phase III trial who received brivanib – an investigational oral inhibitor of signaling in the vascular endothelial growth factor (VEGF) and fibroblast growth factor (FGF) pathways – lived a median of about 9 months, compared with 8 months for those given placebo, Dr. Lillian L. Siu reported at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology.

Adding brivanib did prolong progression-free survival and increase response rate, but the tradeoff was more toxicity, greater treatment discontinuation, and faster deterioration in quality of life, she said.

"There is good scientific rationale to investigate [brivanib] in combination with EGFR [epidermal growth factor receptor] inhibitors" such as cetuximab, commented Dr. Siu, a senior staff physician in the division of medical oncology and hematology at Princess Margaret Hospital, Toronto. The CO.20 trial, however, "did not meet its primary end point of overall survival," she said. Analyses assessing potential biomarkers are still ongoing.

"The median does not well reflect the overall benefit in progression-free survival by the nature of the curves, which tend to splay at the 50-percentile mark," pointed out discussant Dr. Herbert Hurwitz of the Duke Cancer Institute in Durham, N.C.

"The side effect profile for this study was unanticipatedly large," he maintained, and the increased toxicity was "clinically meaningful," leading to higher dropout and to "significant dose reductions in respect to brivanib, but also in respect to cetuximab, thereby compromising not only the value of the experimental drug, but of the standard drug as well." Thus, "the toxicity of this doublet likely confounded some of the efficacy results," he added.

The trial’s findings call into question the role in colorectal cancer of basic FGF (bFGF) as a mechanism of resistance to anti-VEGF therapy, suggested by early research. "Although this [trial] does not exclude a benefit, we need to rethink exactly what is the role, particularly given the diagnostic difficulty of assessing bFGF in both plasma and tissue," Dr. Hurwitz said.

"This data clearly will not lead to a regulatory approval" of brivanib, he concluded. "Data as monotherapy are not known, and data in trials in hepatocellular cancer are still ongoing."

To be eligible for the CO.20 trial, patients were required to have previously received a thymidylate synthase inhibitor (such as 5-fluorouracil or capecitabine) and to be intolerant of or have disease refractory to irinotecan and oxaliplatin. They could have received, at most, one prior therapy targeting VEGF but not any prior therapies targeting EGFR.

In all, 750 patients were assigned evenly to weekly cetuximab (Erbitux), an antibody to EGFR, plus either daily placebo or daily brivanib.

The patients had median age of about 64 years; nearly all (92%) had received more than three prior lines of chemotherapy in any setting (neoadjuvant, adjuvant, and/or metastatic), and a large share (41%) had received a prior therapy targeting VEGF.

Main results showed that, with a median follow-up of 19 months, there was no significant gain in median overall survival with brivanib-cetuximab compared with placebo-cetuximab in intention-to-treat analyses (8.8 vs. 8.1 months). The relative benefit was essentially the same in all patient subgroups.

Brivanib was associated with better progression-free survival (5.0 vs. 3.4 months; hazard ratio, 0.72; P less than .0001), a finding consistent across subgroups, and with a higher response rate (13.6% vs. 7.2%, P = .004).

The proportions of patients receiving at least 90% of the planned dose intensities of brivanib/placebo and cetuximab were lower in the brivanib group. "This was mainly due to dose reductions as well as omissions of these drugs," Dr. Siu noted.

The overall rate of grade 3 or worse nonhematologic toxicity was higher with brivanib than with placebo (78% vs. 53%). Fatigue, hypertension, gastrointestinal toxicity, and certain biochemical abnormalities were all more common with the drug. The rate of discontinuation as a result of adverse events was also higher with brivanib (22%) than with placebo (3%).

Finally, the median time to deterioration of quality of life was shorter with brivanib than with placebo for both global health (1.1 vs. 1.6 months, P = .02) and physical function (1.7 vs. 5.6 months, P less than .0001).

Dr. Siu reported that she receives research funding from Bristol-Myers Squibb. Dr. Hurwitz reported that he is a consultant to Bristol-Myers Squibb and Genentech/Roche; receives honoraria from Roche; and receives research funding from Acceleron Pharma, Amgen, Bristol-Myers Squibb, Cephalon, Genentech, GlaxoSmithKline, Novartis, Pfizer, Roche, Sanofi-Aventis, Threshold, and Tracon.

SAN FRANCISCO – Eagerly awaited results of the CO.20 trial in metastatic chemorefractory KRAS wild-type colorectal cancer are disappointing, showing no significant gain in the primary end point of overall survival from the addition of brivanib to cetuximab.

Patients in the phase III trial who received brivanib – an investigational oral inhibitor of signaling in the vascular endothelial growth factor (VEGF) and fibroblast growth factor (FGF) pathways – lived a median of about 9 months, compared with 8 months for those given placebo, Dr. Lillian L. Siu reported at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology.

Adding brivanib did prolong progression-free survival and increase response rate, but the tradeoff was more toxicity, greater treatment discontinuation, and faster deterioration in quality of life, she said.

"There is good scientific rationale to investigate [brivanib] in combination with EGFR [epidermal growth factor receptor] inhibitors" such as cetuximab, commented Dr. Siu, a senior staff physician in the division of medical oncology and hematology at Princess Margaret Hospital, Toronto. The CO.20 trial, however, "did not meet its primary end point of overall survival," she said. Analyses assessing potential biomarkers are still ongoing.

"The median does not well reflect the overall benefit in progression-free survival by the nature of the curves, which tend to splay at the 50-percentile mark," pointed out discussant Dr. Herbert Hurwitz of the Duke Cancer Institute in Durham, N.C.

"The side effect profile for this study was unanticipatedly large," he maintained, and the increased toxicity was "clinically meaningful," leading to higher dropout and to "significant dose reductions in respect to brivanib, but also in respect to cetuximab, thereby compromising not only the value of the experimental drug, but of the standard drug as well." Thus, "the toxicity of this doublet likely confounded some of the efficacy results," he added.

The trial’s findings call into question the role in colorectal cancer of basic FGF (bFGF) as a mechanism of resistance to anti-VEGF therapy, suggested by early research. "Although this [trial] does not exclude a benefit, we need to rethink exactly what is the role, particularly given the diagnostic difficulty of assessing bFGF in both plasma and tissue," Dr. Hurwitz said.

"This data clearly will not lead to a regulatory approval" of brivanib, he concluded. "Data as monotherapy are not known, and data in trials in hepatocellular cancer are still ongoing."

To be eligible for the CO.20 trial, patients were required to have previously received a thymidylate synthase inhibitor (such as 5-fluorouracil or capecitabine) and to be intolerant of or have disease refractory to irinotecan and oxaliplatin. They could have received, at most, one prior therapy targeting VEGF but not any prior therapies targeting EGFR.

In all, 750 patients were assigned evenly to weekly cetuximab (Erbitux), an antibody to EGFR, plus either daily placebo or daily brivanib.

The patients had median age of about 64 years; nearly all (92%) had received more than three prior lines of chemotherapy in any setting (neoadjuvant, adjuvant, and/or metastatic), and a large share (41%) had received a prior therapy targeting VEGF.

Main results showed that, with a median follow-up of 19 months, there was no significant gain in median overall survival with brivanib-cetuximab compared with placebo-cetuximab in intention-to-treat analyses (8.8 vs. 8.1 months). The relative benefit was essentially the same in all patient subgroups.

Brivanib was associated with better progression-free survival (5.0 vs. 3.4 months; hazard ratio, 0.72; P less than .0001), a finding consistent across subgroups, and with a higher response rate (13.6% vs. 7.2%, P = .004).

The proportions of patients receiving at least 90% of the planned dose intensities of brivanib/placebo and cetuximab were lower in the brivanib group. "This was mainly due to dose reductions as well as omissions of these drugs," Dr. Siu noted.

The overall rate of grade 3 or worse nonhematologic toxicity was higher with brivanib than with placebo (78% vs. 53%). Fatigue, hypertension, gastrointestinal toxicity, and certain biochemical abnormalities were all more common with the drug. The rate of discontinuation as a result of adverse events was also higher with brivanib (22%) than with placebo (3%).

Finally, the median time to deterioration of quality of life was shorter with brivanib than with placebo for both global health (1.1 vs. 1.6 months, P = .02) and physical function (1.7 vs. 5.6 months, P less than .0001).

Dr. Siu reported that she receives research funding from Bristol-Myers Squibb. Dr. Hurwitz reported that he is a consultant to Bristol-Myers Squibb and Genentech/Roche; receives honoraria from Roche; and receives research funding from Acceleron Pharma, Amgen, Bristol-Myers Squibb, Cephalon, Genentech, GlaxoSmithKline, Novartis, Pfizer, Roche, Sanofi-Aventis, Threshold, and Tracon.

SAN FRANCISCO – The benefit of bevacizumab in some patients with advanced gastric cancer appears to vary by disease subtype, based on an exploratory analysis of data from the phase III AVAGAST trial.

The global trial randomized 774 patients to chemotherapy plus placebo or bevacizumab (Avastin), an antibody to vascular endothelial growth factor (VEGF). Results for the entire trial population, which were reported previously (J. Clin. Oncol. 2011;29:3968-76), showed no significant gain in overall survival from the addition of bevacizumab.

In the new analysis, presented at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology, investigators divided the patients according to three types of disease with distinct histology and clinical and environmental risk factors. About 10% of all patients had proximal (also known as type 1) disease, 50% had diffuse (type 2) disease, and 40% had distal (type 3) disease.

The main results showed that the addition of bevacizumab to chemotherapy was associated with a significant, one-third reduction in the risk of death for patients from Europe and the Americas who had diffuse or distal disease. Patients with proximal disease, and patients from the Asia-Pacific region with any disease type, did not show significant benefit.

Additionally, in the trial’s control arm, overall survival in patients from Europe and the Americas differed across the three types, with the diffuse type carrying the poorest prognosis. And levels of two biomarkers showing marginal correlation with bevacizumab efficacy – tumor neuropilin-1 (NRP-1) and plasma VEGF-A – varied across the types in ways that were consistent with the observed drug benefit.

"Gastric cancers really are more than one disease. The subtypes have different prognoses; they have different [bevacizumab] efficacy," said first author Dr. Manish A. Shah, a gastrointestinal oncologist with the Memorial Sloan-Kettering Cancer Center in New York. These subtypes "may form the basis of a new classification schema for gastric cancer."

In addition, the candidate biomarkers tumor NRP-1 and plasma VEGF-A may actually provide a rationale for gastric cancer and/or subtype-specific outcomes with bevacizumab, he said.

Discussant Dr. David H. Ilson agreed that etiologic and molecular data show that "these are likely biologically different diseases. The caveat from Dr. Shah’s analysis, however, is that this is a small, unplanned subset analysis and is at best hypothesis generating."

Numerous Asian trials of chemotherapy in gastric cancer have not found diffuse histology to be either prognostic or predictive, and Western trials have often not considered disease type, Dr. Ilson noted. Thus, "the potential prognostic and predictive impact of diffuse histology needs to be validated in other cohorts. And the potential bevacizumab biomarkers also require validation."

The need for further study of bevacizumab in gastric cancer in the West is debatable, according to Dr. Ilson, also of the Memorial Sloan-Kettering Cancer Center. "And I think an even bigger and overarching question is, Are agents targeting only one pathway really worth investigating? Should we be looking at multitargeted agents," or combining agents?

"The holy grail here, which never gets enough mention, is that we need to identify and validate biomarkers to direct therapy to the appropriate patient," he said.

In the unplanned, exploratory analysis, Dr. Shah and his team found that among patients from Europe and the Americas, the addition of bevacizumab improved overall survival significantly for those having diffuse disease (hazard ratio, 0.68) and marginally for those having distal disease (0.72). When these two groups were combined, the risk of death was reduced by one-third (0.67).

In an analysis of potential biomarkers of bevacizumab efficacy, survival benefit differed by tumor NRP-1 level (P = .06) – with only patients having below-median levels deriving benefit – and by plasma VEGF-A level (P = .07) – with only patients having above-median values deriving benefit.

Proximal tumors generally had higher tumor NRP-1 levels and lower plasma VEGF-A levels. By comparison, diffuse tumors had relatively low NRP-1 levels, and distal tumors had relatively high plasma VEGF-A levels.

"There is heterogeneity with these biomarkers across the different subtypes, and this is in support of the concept that subtypes are clinically meaningful," Dr. Shah commented. "Proximal tumors tend to have the worst profile, if you will, for bevacizumab."

Dr. Shah reported that he is a consultant to Genentech. Dr. Ilson reported that he receives research funding from Bayer, BMS-Imclone, Roche-Genentech, and Sanofi-Aventis.

SAN FRANCISCO – The benefit of bevacizumab in some patients with advanced gastric cancer appears to vary by disease subtype, based on an exploratory analysis of data from the phase III AVAGAST trial.

The global trial randomized 774 patients to chemotherapy plus placebo or bevacizumab (Avastin), an antibody to vascular endothelial growth factor (VEGF). Results for the entire trial population, which were reported previously (J. Clin. Oncol. 2011;29:3968-76), showed no significant gain in overall survival from the addition of bevacizumab.

In the new analysis, presented at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology, investigators divided the patients according to three types of disease with distinct histology and clinical and environmental risk factors. About 10% of all patients had proximal (also known as type 1) disease, 50% had diffuse (type 2) disease, and 40% had distal (type 3) disease.

The main results showed that the addition of bevacizumab to chemotherapy was associated with a significant, one-third reduction in the risk of death for patients from Europe and the Americas who had diffuse or distal disease. Patients with proximal disease, and patients from the Asia-Pacific region with any disease type, did not show significant benefit.

Additionally, in the trial’s control arm, overall survival in patients from Europe and the Americas differed across the three types, with the diffuse type carrying the poorest prognosis. And levels of two biomarkers showing marginal correlation with bevacizumab efficacy – tumor neuropilin-1 (NRP-1) and plasma VEGF-A – varied across the types in ways that were consistent with the observed drug benefit.

"Gastric cancers really are more than one disease. The subtypes have different prognoses; they have different [bevacizumab] efficacy," said first author Dr. Manish A. Shah, a gastrointestinal oncologist with the Memorial Sloan-Kettering Cancer Center in New York. These subtypes "may form the basis of a new classification schema for gastric cancer."

In addition, the candidate biomarkers tumor NRP-1 and plasma VEGF-A may actually provide a rationale for gastric cancer and/or subtype-specific outcomes with bevacizumab, he said.

Discussant Dr. David H. Ilson agreed that etiologic and molecular data show that "these are likely biologically different diseases. The caveat from Dr. Shah’s analysis, however, is that this is a small, unplanned subset analysis and is at best hypothesis generating."

Numerous Asian trials of chemotherapy in gastric cancer have not found diffuse histology to be either prognostic or predictive, and Western trials have often not considered disease type, Dr. Ilson noted. Thus, "the potential prognostic and predictive impact of diffuse histology needs to be validated in other cohorts. And the potential bevacizumab biomarkers also require validation."

The need for further study of bevacizumab in gastric cancer in the West is debatable, according to Dr. Ilson, also of the Memorial Sloan-Kettering Cancer Center. "And I think an even bigger and overarching question is, Are agents targeting only one pathway really worth investigating? Should we be looking at multitargeted agents," or combining agents?

"The holy grail here, which never gets enough mention, is that we need to identify and validate biomarkers to direct therapy to the appropriate patient," he said.

In the unplanned, exploratory analysis, Dr. Shah and his team found that among patients from Europe and the Americas, the addition of bevacizumab improved overall survival significantly for those having diffuse disease (hazard ratio, 0.68) and marginally for those having distal disease (0.72). When these two groups were combined, the risk of death was reduced by one-third (0.67).

In an analysis of potential biomarkers of bevacizumab efficacy, survival benefit differed by tumor NRP-1 level (P = .06) – with only patients having below-median levels deriving benefit – and by plasma VEGF-A level (P = .07) – with only patients having above-median values deriving benefit.

Proximal tumors generally had higher tumor NRP-1 levels and lower plasma VEGF-A levels. By comparison, diffuse tumors had relatively low NRP-1 levels, and distal tumors had relatively high plasma VEGF-A levels.

"There is heterogeneity with these biomarkers across the different subtypes, and this is in support of the concept that subtypes are clinically meaningful," Dr. Shah commented. "Proximal tumors tend to have the worst profile, if you will, for bevacizumab."

Dr. Shah reported that he is a consultant to Genentech. Dr. Ilson reported that he receives research funding from Bayer, BMS-Imclone, Roche-Genentech, and Sanofi-Aventis.

SAN FRANCISCO – The benefit of bevacizumab in some patients with advanced gastric cancer appears to vary by disease subtype, based on an exploratory analysis of data from the phase III AVAGAST trial.

The global trial randomized 774 patients to chemotherapy plus placebo or bevacizumab (Avastin), an antibody to vascular endothelial growth factor (VEGF). Results for the entire trial population, which were reported previously (J. Clin. Oncol. 2011;29:3968-76), showed no significant gain in overall survival from the addition of bevacizumab.

In the new analysis, presented at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology, investigators divided the patients according to three types of disease with distinct histology and clinical and environmental risk factors. About 10% of all patients had proximal (also known as type 1) disease, 50% had diffuse (type 2) disease, and 40% had distal (type 3) disease.

The main results showed that the addition of bevacizumab to chemotherapy was associated with a significant, one-third reduction in the risk of death for patients from Europe and the Americas who had diffuse or distal disease. Patients with proximal disease, and patients from the Asia-Pacific region with any disease type, did not show significant benefit.

Additionally, in the trial’s control arm, overall survival in patients from Europe and the Americas differed across the three types, with the diffuse type carrying the poorest prognosis. And levels of two biomarkers showing marginal correlation with bevacizumab efficacy – tumor neuropilin-1 (NRP-1) and plasma VEGF-A – varied across the types in ways that were consistent with the observed drug benefit.

"Gastric cancers really are more than one disease. The subtypes have different prognoses; they have different [bevacizumab] efficacy," said first author Dr. Manish A. Shah, a gastrointestinal oncologist with the Memorial Sloan-Kettering Cancer Center in New York. These subtypes "may form the basis of a new classification schema for gastric cancer."

In addition, the candidate biomarkers tumor NRP-1 and plasma VEGF-A may actually provide a rationale for gastric cancer and/or subtype-specific outcomes with bevacizumab, he said.

Discussant Dr. David H. Ilson agreed that etiologic and molecular data show that "these are likely biologically different diseases. The caveat from Dr. Shah’s analysis, however, is that this is a small, unplanned subset analysis and is at best hypothesis generating."

Numerous Asian trials of chemotherapy in gastric cancer have not found diffuse histology to be either prognostic or predictive, and Western trials have often not considered disease type, Dr. Ilson noted. Thus, "the potential prognostic and predictive impact of diffuse histology needs to be validated in other cohorts. And the potential bevacizumab biomarkers also require validation."

The need for further study of bevacizumab in gastric cancer in the West is debatable, according to Dr. Ilson, also of the Memorial Sloan-Kettering Cancer Center. "And I think an even bigger and overarching question is, Are agents targeting only one pathway really worth investigating? Should we be looking at multitargeted agents," or combining agents?

"The holy grail here, which never gets enough mention, is that we need to identify and validate biomarkers to direct therapy to the appropriate patient," he said.

In the unplanned, exploratory analysis, Dr. Shah and his team found that among patients from Europe and the Americas, the addition of bevacizumab improved overall survival significantly for those having diffuse disease (hazard ratio, 0.68) and marginally for those having distal disease (0.72). When these two groups were combined, the risk of death was reduced by one-third (0.67).

In an analysis of potential biomarkers of bevacizumab efficacy, survival benefit differed by tumor NRP-1 level (P = .06) – with only patients having below-median levels deriving benefit – and by plasma VEGF-A level (P = .07) – with only patients having above-median values deriving benefit.

Proximal tumors generally had higher tumor NRP-1 levels and lower plasma VEGF-A levels. By comparison, diffuse tumors had relatively low NRP-1 levels, and distal tumors had relatively high plasma VEGF-A levels.

"There is heterogeneity with these biomarkers across the different subtypes, and this is in support of the concept that subtypes are clinically meaningful," Dr. Shah commented. "Proximal tumors tend to have the worst profile, if you will, for bevacizumab."

Dr. Shah reported that he is a consultant to Genentech. Dr. Ilson reported that he receives research funding from Bayer, BMS-Imclone, Roche-Genentech, and Sanofi-Aventis.

SAN FRANCISCO – None of the common KRAS mutant alleles seen in metastatic colorectal cancer reliably predict benefit from panitumumab, new data show, suggesting that this agent should continue to be used only in patients with KRAS wild-type tumors.

Investigators led by Dr. Marc Peeters of Antwerp University Hospital in Edegem, Belgium, analyzed data from a trio of phase III randomized trials of panitumumab (Vectibix), an antibody to epidermal growth factor receptor (EGFR), in more than 1,400 patients.

Main results showed that there was no consistent difference in benefit across trials in terms of progression-free and overall survival for patients having the most common codon 12 and 13 KRAS mutant alleles, he reported at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology.

In a pooled analysis, a significant association was seen only for the G12A mutant allele, whereby patients having this allele had poorer overall survival when panitumumab was added to their treatment.

"The take-home message based on this study: Patients with metastatic colorectal cancer harboring any of the most common codon 12 or 13 mutant KRAS alleles are unlikely to benefit from panitumumab treatment. Today, only metastatic colorectal cancer patients with wild-type KRAS tumors should be treated with EGFR antibodies," Dr. Peeters asserted.

Previous research (JAMA 2010;304:1812-20), has suggested that patients with the G13D KRAS mutant allele might still benefit from cetuximab, another antibody to EGFR, an attendee noted. Therefore, is it possible that there is some biological difference between panitumumab and cetuximab (Erbitux) when it comes to this allele?

"The results I presented here were based on a large data set treated with panitumumab and a very homogeneous population, which really shows ... that there is no benefit of treating patients with G13D with panitumumab," Dr. Peeters replied. "If you compare the results ... of data with cetuximab, this population is more heterogeneous than we studied in the panitumumab group.

"So it’s very difficult to draw final conclusions ... on standard of care of patients. My advice is still that patients with mutant tumors probably in general don’t benefit from treatment with EGFR antibodies, although from a scientific point of view, the discussion is still open and we have to search further [to see] if there is really a difference between the antibodies based on the results obtained in the literature," he added.

"And one of the studies that could answer this question is a head-to-head study that we are performing, where cetuximab is compared with panitumumab, and this might give us a better idea what is really the difference between the antibodies."

Session cochair Dr. George Fisher of the Stanford (Calif.) Cancer Center, maintained that the "door has been closed now" on potential benefit of anti-EGFR therapy in subsets of patients with KRAS-mutated colorectal cancer. "It’s hard for me to imagine that any mutations short of the very rare ones ... might be appropriate for EGFR inhibitors," he said in an interview.

However, even if inhibition of signaling in the EGFR pathway in these patients is not efficacious, there still might be a role for EGFR binding as it relates to antibody-dependent cell-mediated cytotoxicity, Dr. Fisher added. "There are a number of groups still working on that, and that would theoretically work even in KRAS mutants."

Panitumumab is approved by the Food and Drug Administration for the treatment of metastatic colorectal cancer. Its label indicates that it is not recommended for patients having common KRAS mutations.

The three trials included in the analysis tested addition of panitumumab to FOLFOX4 in the first-line setting (the PRIME trial, 301 patients), to FOLFIRI in the second-line setting (701 patients), and to best supportive care in the third-line setting (463 patients). In the new analysis, the investigators assessed the prognostic and predictive impact of six common KRAS mutant alleles: G12D, G12V, G13D, G12C, G12A, and G12S.

"Baseline demographics and clinical features were generally balanced between all mutant KRAS allele subgroups," Dr. Peeters noted.

In cross-trial analyses in which researchers looked just at the control arms to assess prognostic impact of the mutant alleles, none of the six alleles were consistently associated with better progression-free survival or overall survival, compared with all other mutant alleles.

When researchers looked at both arms in the trials to assess predictive impact of the mutant alleles, none of them were consistently associated with differential progression-free survival or overall survival if patients received added panitumumab instead of the control treatment.

In a single trial, G12V was associated with better overall survival and G13D was associated with poorer overall survival if patients received added panitumumab instead of the control treatment of FOLFOX4 alone, with a significant interaction term seen for each allele (P = .037 and P = .002, respectively).

In a pooled analysis of all three trials, only one mutant allele (G12) was significantly associated with overall survival, with poorer overall survival seen for patients having this allele if they received added panitumumab instead of the control treatment alone.

(The investigators also evaluated data from the CRYSTAL and OPUS trials, which tested addition of cetuximab to chemotherapy, and found that neither the G12V nor the G13D mutant allele significantly predicted benefit from the antibody in these trials.)

Dr. Peeters reported that he is a consultant to and receives honoraria and research funding from Amgen. Dr. Fisher reported that he owns stock in Seattle Genetics and receives research funding from Exelixis, FibroGen, Genentech, GlaxoSmithKline, Ispen, Merck, Newlink Genetics, Novartis, and Sanofi Aventis.

SAN FRANCISCO – None of the common KRAS mutant alleles seen in metastatic colorectal cancer reliably predict benefit from panitumumab, new data show, suggesting that this agent should continue to be used only in patients with KRAS wild-type tumors.

Investigators led by Dr. Marc Peeters of Antwerp University Hospital in Edegem, Belgium, analyzed data from a trio of phase III randomized trials of panitumumab (Vectibix), an antibody to epidermal growth factor receptor (EGFR), in more than 1,400 patients.

Main results showed that there was no consistent difference in benefit across trials in terms of progression-free and overall survival for patients having the most common codon 12 and 13 KRAS mutant alleles, he reported at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology.

In a pooled analysis, a significant association was seen only for the G12A mutant allele, whereby patients having this allele had poorer overall survival when panitumumab was added to their treatment.

"The take-home message based on this study: Patients with metastatic colorectal cancer harboring any of the most common codon 12 or 13 mutant KRAS alleles are unlikely to benefit from panitumumab treatment. Today, only metastatic colorectal cancer patients with wild-type KRAS tumors should be treated with EGFR antibodies," Dr. Peeters asserted.

Previous research (JAMA 2010;304:1812-20), has suggested that patients with the G13D KRAS mutant allele might still benefit from cetuximab, another antibody to EGFR, an attendee noted. Therefore, is it possible that there is some biological difference between panitumumab and cetuximab (Erbitux) when it comes to this allele?

"The results I presented here were based on a large data set treated with panitumumab and a very homogeneous population, which really shows ... that there is no benefit of treating patients with G13D with panitumumab," Dr. Peeters replied. "If you compare the results ... of data with cetuximab, this population is more heterogeneous than we studied in the panitumumab group.

"So it’s very difficult to draw final conclusions ... on standard of care of patients. My advice is still that patients with mutant tumors probably in general don’t benefit from treatment with EGFR antibodies, although from a scientific point of view, the discussion is still open and we have to search further [to see] if there is really a difference between the antibodies based on the results obtained in the literature," he added.

"And one of the studies that could answer this question is a head-to-head study that we are performing, where cetuximab is compared with panitumumab, and this might give us a better idea what is really the difference between the antibodies."

Session cochair Dr. George Fisher of the Stanford (Calif.) Cancer Center, maintained that the "door has been closed now" on potential benefit of anti-EGFR therapy in subsets of patients with KRAS-mutated colorectal cancer. "It’s hard for me to imagine that any mutations short of the very rare ones ... might be appropriate for EGFR inhibitors," he said in an interview.

However, even if inhibition of signaling in the EGFR pathway in these patients is not efficacious, there still might be a role for EGFR binding as it relates to antibody-dependent cell-mediated cytotoxicity, Dr. Fisher added. "There are a number of groups still working on that, and that would theoretically work even in KRAS mutants."

Panitumumab is approved by the Food and Drug Administration for the treatment of metastatic colorectal cancer. Its label indicates that it is not recommended for patients having common KRAS mutations.

The three trials included in the analysis tested addition of panitumumab to FOLFOX4 in the first-line setting (the PRIME trial, 301 patients), to FOLFIRI in the second-line setting (701 patients), and to best supportive care in the third-line setting (463 patients). In the new analysis, the investigators assessed the prognostic and predictive impact of six common KRAS mutant alleles: G12D, G12V, G13D, G12C, G12A, and G12S.

"Baseline demographics and clinical features were generally balanced between all mutant KRAS allele subgroups," Dr. Peeters noted.

In cross-trial analyses in which researchers looked just at the control arms to assess prognostic impact of the mutant alleles, none of the six alleles were consistently associated with better progression-free survival or overall survival, compared with all other mutant alleles.

When researchers looked at both arms in the trials to assess predictive impact of the mutant alleles, none of them were consistently associated with differential progression-free survival or overall survival if patients received added panitumumab instead of the control treatment.

In a single trial, G12V was associated with better overall survival and G13D was associated with poorer overall survival if patients received added panitumumab instead of the control treatment of FOLFOX4 alone, with a significant interaction term seen for each allele (P = .037 and P = .002, respectively).

In a pooled analysis of all three trials, only one mutant allele (G12) was significantly associated with overall survival, with poorer overall survival seen for patients having this allele if they received added panitumumab instead of the control treatment alone.

(The investigators also evaluated data from the CRYSTAL and OPUS trials, which tested addition of cetuximab to chemotherapy, and found that neither the G12V nor the G13D mutant allele significantly predicted benefit from the antibody in these trials.)

Dr. Peeters reported that he is a consultant to and receives honoraria and research funding from Amgen. Dr. Fisher reported that he owns stock in Seattle Genetics and receives research funding from Exelixis, FibroGen, Genentech, GlaxoSmithKline, Ispen, Merck, Newlink Genetics, Novartis, and Sanofi Aventis.

SAN FRANCISCO – None of the common KRAS mutant alleles seen in metastatic colorectal cancer reliably predict benefit from panitumumab, new data show, suggesting that this agent should continue to be used only in patients with KRAS wild-type tumors.

Investigators led by Dr. Marc Peeters of Antwerp University Hospital in Edegem, Belgium, analyzed data from a trio of phase III randomized trials of panitumumab (Vectibix), an antibody to epidermal growth factor receptor (EGFR), in more than 1,400 patients.

Main results showed that there was no consistent difference in benefit across trials in terms of progression-free and overall survival for patients having the most common codon 12 and 13 KRAS mutant alleles, he reported at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology.

In a pooled analysis, a significant association was seen only for the G12A mutant allele, whereby patients having this allele had poorer overall survival when panitumumab was added to their treatment.

"The take-home message based on this study: Patients with metastatic colorectal cancer harboring any of the most common codon 12 or 13 mutant KRAS alleles are unlikely to benefit from panitumumab treatment. Today, only metastatic colorectal cancer patients with wild-type KRAS tumors should be treated with EGFR antibodies," Dr. Peeters asserted.

Previous research (JAMA 2010;304:1812-20), has suggested that patients with the G13D KRAS mutant allele might still benefit from cetuximab, another antibody to EGFR, an attendee noted. Therefore, is it possible that there is some biological difference between panitumumab and cetuximab (Erbitux) when it comes to this allele?

"The results I presented here were based on a large data set treated with panitumumab and a very homogeneous population, which really shows ... that there is no benefit of treating patients with G13D with panitumumab," Dr. Peeters replied. "If you compare the results ... of data with cetuximab, this population is more heterogeneous than we studied in the panitumumab group.

"So it’s very difficult to draw final conclusions ... on standard of care of patients. My advice is still that patients with mutant tumors probably in general don’t benefit from treatment with EGFR antibodies, although from a scientific point of view, the discussion is still open and we have to search further [to see] if there is really a difference between the antibodies based on the results obtained in the literature," he added.

"And one of the studies that could answer this question is a head-to-head study that we are performing, where cetuximab is compared with panitumumab, and this might give us a better idea what is really the difference between the antibodies."

Session cochair Dr. George Fisher of the Stanford (Calif.) Cancer Center, maintained that the "door has been closed now" on potential benefit of anti-EGFR therapy in subsets of patients with KRAS-mutated colorectal cancer. "It’s hard for me to imagine that any mutations short of the very rare ones ... might be appropriate for EGFR inhibitors," he said in an interview.

However, even if inhibition of signaling in the EGFR pathway in these patients is not efficacious, there still might be a role for EGFR binding as it relates to antibody-dependent cell-mediated cytotoxicity, Dr. Fisher added. "There are a number of groups still working on that, and that would theoretically work even in KRAS mutants."

Panitumumab is approved by the Food and Drug Administration for the treatment of metastatic colorectal cancer. Its label indicates that it is not recommended for patients having common KRAS mutations.

The three trials included in the analysis tested addition of panitumumab to FOLFOX4 in the first-line setting (the PRIME trial, 301 patients), to FOLFIRI in the second-line setting (701 patients), and to best supportive care in the third-line setting (463 patients). In the new analysis, the investigators assessed the prognostic and predictive impact of six common KRAS mutant alleles: G12D, G12V, G13D, G12C, G12A, and G12S.

"Baseline demographics and clinical features were generally balanced between all mutant KRAS allele subgroups," Dr. Peeters noted.

In cross-trial analyses in which researchers looked just at the control arms to assess prognostic impact of the mutant alleles, none of the six alleles were consistently associated with better progression-free survival or overall survival, compared with all other mutant alleles.

When researchers looked at both arms in the trials to assess predictive impact of the mutant alleles, none of them were consistently associated with differential progression-free survival or overall survival if patients received added panitumumab instead of the control treatment.

In a single trial, G12V was associated with better overall survival and G13D was associated with poorer overall survival if patients received added panitumumab instead of the control treatment of FOLFOX4 alone, with a significant interaction term seen for each allele (P = .037 and P = .002, respectively).

In a pooled analysis of all three trials, only one mutant allele (G12) was significantly associated with overall survival, with poorer overall survival seen for patients having this allele if they received added panitumumab instead of the control treatment alone.

(The investigators also evaluated data from the CRYSTAL and OPUS trials, which tested addition of cetuximab to chemotherapy, and found that neither the G12V nor the G13D mutant allele significantly predicted benefit from the antibody in these trials.)

Dr. Peeters reported that he is a consultant to and receives honoraria and research funding from Amgen. Dr. Fisher reported that he owns stock in Seattle Genetics and receives research funding from Exelixis, FibroGen, Genentech, GlaxoSmithKline, Ispen, Merck, Newlink Genetics, Novartis, and Sanofi Aventis.