SAN FRANCISCO-Retrospective analyses have demonstrated a strong correlation between high expression of the DNA-repair gene ERCC1 in tumors and poor response to platinum-based chemotherapy in a variety of cancers. Now, a correlative study associated with Southwest Oncology Group (SWOG) trial 0356 supports that link in esophageal cancer, researchers reported here at the Gastrointestinal Cancers Symposium.
As a result of these data, SWOG is planning a trial in which patients will be assigned to treatment based on the level of ERCC1 expression in their tumor.
Still, however, not everyone thinks the biomarker has been validated sufficiently to guide treatment, even in a clinical trial.
"Our study [Abstract 2] was designed to validate, prospectively, ERCC1 gene expression as a biomarker predicting outcome in patients treated with oxaliplatin-based chemotherapy in combination with radiation in the SWOG 0536 trial," said Pierre O. Bohanes, MD, an oncology fellow at the University of Southern California Norris Comprehensive Cancer Center, who presented the data.
"When we looked at progression-free survival by ERCC1 RNA levels, using the predefined cutoff, there was a higher risk of recurrence in the high-ERCC1 group, with the hazard ratio nearly reaching three."
Dr. Bohanes was given a Merit Award for Outstanding Research at the meeting, an honor recognizing the work of oncology fellows who are first authors on abstracts considered to be outstanding.
Required for Nucleotide Excision Repair
ERCC1 is required for nucleotide excision repair-i.e., the DNA repair mechanism that detects and removes damage caused by platinum agents. Preclinical and retrospective studies suggest that ERCC1 expression level may be a useful biomarker for identifying patients who would benefit from non-platinum chemotherapy in a variety of cancers, including non-small cell lung, gastric, and head and neck cancers.
Additionally, one prospective trial in NSCLC showed that selecting patient therapy based on ERCC1 expression level can increase overall response rates in the low-ERCC1 group (Cobo et al: JOCK 2007;25:2747-2754).
To prospectively test the biomarker's value in a gastrointestinal cancer, Dr. Bohanes and colleagues compared ERCC1 expression with clinical outcomes in patients enrolled in SWOG 0356. The single-arm Phase II trial tested oxaliplatin-based chemotherapy in patients with clinical Stage II or III esophageal or gastroesophageal junction and enrolled 98 patients between February 2005 and August 2008.
Of those, 92 were eligible for outcome assessment and the biomarker correlative study. Tumor samples were obtained from 90 of those patients, but sufficient material to test ERCC1 RNA levels was available in only 53 samples.
There were no substantial differences between the patients in this subgroup and the overall study population, Dr. Bohanes said. In the entire study population, 28.2% of patients achieved a pathological complete response and 11% had minimal residual disease. With a median follow-up of 36.8 months, median progression-free survival was 20.8 months and median overall survival was 33.7 months.
Patients whose tumors expressed a low level of ERCC1 gained the most benefit from the therapy. Based on a predefined cutoff, 22 patients had low ERCC1 gene expression and 31 had high expression. Median progression-free survival for patients with low ERCC1 was not reached with a median follow-up of 36.8 months, compared with 14.8 months for those with high ERCC1, with a hazard ratio or 2.97.
The two-year progression-free survival rates were 67% and 17% in the low and high groups, respectively. Median overall survival was not reached in the low-ERCC1 group compared with 22.4 months in the high-ERCC1 groups, with a hazard ratio of 2.32.
The two-year overall survival was 72% and 37% in the low and high groups, respectively. ERCC1 levels, however, were not associated with pathological complete response.
The investigators also analyzed the expression of six other genes involved in DNA repair, platinum detoxification or fluorouracil metabolism, but found that none were associated with clinical outcomes. Similarly, none of the 13 genetic polymorphisms found in patients germ-line DNA were associated with outcome.
Dr. Bohanes noted three study limitations during his presentation, including the relatively small sample size, insufficient tumor samples from many patients, and no differentiation between clinical Stage II and clinical Stage III patients.
Concerns about Lack of Clinical Staging Information
It was the last point that raised the most concern from other experts. For example, David H. Ilson, MD, PhD, Attending Physician in the Gastrointestinal Oncology Service at Memorial Sloan-Kettering Cancer Center, noted that while it was impressive to do such correlative work in a cooperative group setting, the lack of clinical staging information was a major problem because ERCC1 expression may not be independent of stage.
"What if it turned out that all of the early-stage patients had low ERCC1? You have no idea if it is an independent factor," he said.
Heinz-Josef Lenz, MD, Associate Director of the Norris Cancer Center Gastrointestinal Oncology Program and Associate Professor of Medicine at Keck School of Medicine of the University of Southern California in Los Angeles, who was senior author on the study and had presented an overview on predictive and prognostic markers in esophageal-gastric cancer earlier in the session, smiled as he acknowledged Dr. Ilson's point: "Obviously you picked out the weakness of this study. We do not know if there is a correlation with TNM stage in this disease," Dr. Lenz said.
"For colon cancer, ERCC1 expression was not related to stage. But you are right-we have no control."
Dr. Lenz continued on to say that the trial organizers had purposely chosen not to require endoscopic ultrasound, which would have given them staging information, as part of the design so as not to exclude community physicians, who may not have such technology available.
The other point Dr. Ilson made both in the session and in a subsequent interview, is that there is at least one retrospective analysis that suggests that patients with high ERCC1 do better with surgery only compared with patients with low ERCC1. If that is the case, then treating these patients with an oxaliplatin-based chemotherapy could worsen their outcomes.
Again, Dr. Lenz said that Dr. Ilson's point was correct but that the retrospective data that Dr. Ilson referred to was from a single study. "We need more validation to build on it as prognostic marker," Dr. Lenz said. "There is a hint, but I would not use it at this time to say 'no treatment.'"
Next Steps
Despite these scientifically valid concerns, Dr. Lenz emphasized that he thinks a prospective biomarker-driven trial that assigns patient treatment based on ERCC1 expression makes sense. "I think the next step should be for patients with high ERCC1, who are not benefiting from this treatment, to be treated with other treatments you have developed at your institution," he said to Dr. Ilson. "We have choices, and I think that should be part of the new design concept moving forward."
Another audience member agreed that it was time to test ERCC1 in a biomarker-driven trial, but expressed concern about a straight randomized design in which patients were assigned to either standard treatment or treatment based on biomarker expression. Given the difficultly recruiting patients to such trials in the United States in this disease, it would take "15 years and three sequential Phase III trials" to gain an answer, he said, suggesting instead that an adaptive trial design, such as that in the lung cancer BATTLE trial at the University of Texas MD Anderson Cancer Center, might be more interesting to patients and lead to an answer sooner.
"I could not agree more with your concerns," Dr. Lenz responded. "We are trying but there is not a simple solution. There is lots of discussion going on about how to utilize our full technology, as well as clinical trial designs, to make bigger steps faster. But I don't think I have a really good answer yet, except that we are incorporating more technologies and flexible designs that adapt to new findings so that we are not stuck with a trial for the next three or five years because of rules."
'Doing Exactly the Right Thing'
After the session, Mitchell C. Posner, MD, the Thomas D. Jones Professor of Surgery and Chief of Surgical Oncology at the University of Chicago Medical Center, told OT that he thought the work was very good. "I think what they are doing is exactly the right thing," Dr. Posner said. "I really do. It would be nice, in a cooperative group setting, to be able to get tissue on everybody to get larger numbers and, therefore, make it more meaningful, but I think they are doing exactly the right thing."
Patients whose tumors expressed a low level of ERCC1 gained the most benefit from the therapy.
As a result of these data, SWOG is planning a trial in which patients will be assigned to treatment based on the level of ERCC1 expression in their tumor. Not everyone, however, thinks the biomarker has been validated sufficiently to guide treatment, even in a clinical trial.
Meeting Cosponsors
The Gastrointestinal Cancers Symposium is cosponsored by the American Society of Clinical Oncology, the American Gastroenterological Association Institute, the American Society for Radiation Oncology, and the Society of Surgical Oncology.