Skip to main content
Premium Trial:

Request an Annual Quote

New VEGF Biomarker Data Illustrates Complexities of Personalizing Anti-Angiogenic Drugs


By Turna Ray

Data presented this week at a European cancer conference suggests that the association between plasma VEGF-A protein levels and patient response to Genentech's VEGF inhibitor Avastin may be limited to certain types of cancer.

At the European Multidisciplinary Cancer Congress in Stockholm, researchers presented data showing that plasma VEGF-A levels similarly predicted the response of pancreatic and gastric cancer patients to Avastin. The association identified in these disease states was found using an ELISA assay that was designed to gauge shorter isoforms of VEGF-A.

However, when the same assay was used to investigate whether there was a correlation between VEGF-A plasma levels and Avastin treatment in non-small cell lung cancer, metastatic colorectal cancer, and renal cell carcinoma, the marker was found to be prognostic of how the cancer might progress in patients, but did not predict whether they would benefit from Avastin treatment.

It is unknown how these biomarker findings might impact Genentech's case with the US Food and Drug Administration to keep the drug on the market for some breast cancer patients who are responding to the drug.

The Roche subsidiary is currently awaiting a decision from the FDA as to whether the agency will remove Avastin from the market as a breast cancer treatment. In appealing this move, Genentech has asked the FDA to keep the drug available for those who are deriving a clinical benefit from it while the company performs a clinical trial to pin down the molecular features of so-called Avastin super-responders.

The proposed trial would test whether breast cancer patients with elevated VEGF-A plasma levels have a better response than patients with low levels of the protein — a hypothesis born from data from the AVADO study presented at the San Antonio Breast Cancer Symposium in 2010. This data showed that patients with high levels of VEGF-A had a progression-free survival hazard ratio of 0.49, while patients with low levels had a hazard ratio of 0.86. "This finding suggests that patients with high levels of VEGF-A may be more likely to derive a more substantial benefit from Avastin," Genentech said in its appeal to the FDA.

It is unknown if the association between high plasma VEGF-A and improved outcomes to Avastin treatment in two disease settings — pancreatic and gastric cancer — would be enough to sway the FDA about the biomarker as a means for keeping the drug available for a subpopulation of breast cancer patients.

If anything, the findings presented at EMCC about the limited association between plasma VEGF-A and Avastin response in NSCLC, metastatic colorectal cancer, and renal cell cancer patients illustrates the difficult task ahead for researchers who seek to identify valid biomarkers that can be used to personalize treatment with anti-angiogenic drugs.

"We've done a significant amount of research into potential Avastin biomarkers in the past decade and several abstracts were presented at the EMCC meeting this past weekend including data on VEGF-A and … SNPs across a variety of trials of Avastin in advanced cancer," a Genentech spokesperson told PGx Reporter in an e-mail.

"The data to date suggest that high plasma VEGF-A levels in the blood may be associated with improved Avastin efficacy in patients with metastatic breast (as previously reported at SABCS in 2010), gastric, and pancreatic cancers. The same effect has not been observed in analyses of Avastin trials of advanced/metastatic non-small cell lung cancer, renal cell carcinoma, and colorectal cancer," the spokesperson wrote.

Although the lack of an association in these latter disease settings may have been due to differences in sample collection and storage, the variable findings "highlight the complicated nature of angiogenesis," the spokesperson added. "At this stage, it is unknown if these differences are due to sample handling or different tumor biology."

If the FDA allows Genentech to keep marketing Avastin to certain subsets of patients while it conducts a biomarker-based study, the drug developer plans to start recruiting patients in the first quarter of next year, and the study is expected to take three and a half years to complete.


In data from the AVAGAST study presented at EMCC, researchers led by Manish Shah of Memorial Sloan-Kettering Cancer Center randomized 774 gastric cancer patients to receive either six cycles of cisplatin plus capecitabine plus Avastin or placebo until they progressed. Patients enrolled in the study had not been previously treated for inoperable, locally advanced, or metastatic gastric cancer or gastro-esophageal adenocarcinoma.

In this study, there wasn't a significant difference between the two arms in terms of overall survival, according to the abstract. However, there was a statistically significant difference in progression-free survival for patients receiving the Avastin-containing cocktail and those receiving placebo.

Shah et al. found that the patients' outcomes differed by region, with patients described as European and Pan-American deriving more benefit from treatment than Asian patients. To gain a better sense of best responders, the researchers investigated whether plasma VEGF-A levels, which they had available for 712 patients, were predictive of overall survival or progression-free survival.

[ pagebreak ]

This molecular analysis revealed that expression of plasma VEGF-A differed across regions, but was higher in non-Asian patients. Plasma VEGF-A was found to be prognostic of disease progression, since in the placebo group those with high plasma VEGF-A levels had shorter overall survival than those with low levels. Shah et al. also reported that patients with high VEGF-A levels had "a trend towards better effect on progression-free survival or overall survival." This effect was not seen in patients of Asian origin.

The progression-free survival hazard ratio for patients with high VEGF-A was 0.64 compared to 0.89 in those with low plasma levels. Non-Asians with high VEGF-A had a hazard ratio of 0.54 compared to 0.83 for those with low levels. In the Asian patient subset, the hazard ratio for the high group was 0.81 compared to 0.91 for the low VEGF-A plasma group.

"Plasma VEGF-A shows potential as a prognostic and/or predictive biomarker candidate for progression-free survival and overall survival in Avastin-treated [advanced gastric adenocarcinoma] patients, mainly driven by non-Asian patients," Shah et al. concluded. "The current data are similar to observations for plasma VEGF-A from two other independent analyses in metastatic breast cancer (AVADO) and pancreatic cancer (AVITA)."


In a retrospective analysis of the AVITA Phase III trial presented at EMCC, researchers led by Eric Van Cutsem of the University Hospital Gasthuisberg investigated further the relationship between plasma VEGF-A levels and outcomes in metastatic pancreatic cancer patients.

The original AVITA study randomized 607 pancreatic cancer patients to receive either gemcitabine, Tarceva and Avastin, or placebo. Once patients on the placebo arm had disease progression, they were given gemcitabine and Tarceva. In the overall population, patients' survival wasn't impacted based on whether they were receiving Avastin in addition to the Tarceva/gemcitabine regimen. However, researchers did find a statistically significant difference in progression-free survival between the two arms.

As part of this study, plasma samples were collected from 225 patients when they first enrolled in the study. Researchers initially analyzed these samples for four biomarkers using a multiplex ELISA assay, including markers in VEGF-A and VEGFR2. Researchers then analyzed 10 additional angiogenic markers using Aushon Biosystems' SearchLight array-based assay.

In the study, Van Cutsem et al. reported that patients in the biomarker group in the placebo arm had shorter overall survival than the overall population. Patients who received the Avastin-containing regimen and had high baseline levels of plasma VEGF-A experienced better progression-free survival and overall survival than those who were in the placebo arm. Researchers reported similar correlations between plasma VEGFR2 levels and overall survival.

"In this subset analysis, plasma VEGF-A and plasma VEGFR2 were identified as promising biomarker candidates for predicting progression-free survival and overall survival with [Avastin] in patients with metastatic pancreatic cancer," the researchers concluded. "These data confirm the potential predictive value of plasma VEGF-A and plasma VEGFR2 already observed in metastatic breast cancer."

Lacking Correlation

In another study presented at EMCC, researchers led by Gordon Jayson of Christie Hospital and the University of Manchester used an ELISA assay to retest baseline patient samples from three previously conducted studies involving Avastin, including 398 metastatic colorectal cancer patients in the AVF2107g trial; 859 NSCLC patients in the AVAiL trial; and 404 renal cell carcinoma patients in the AVOREN trial.

In this study, Jayson and colleagues did not find the same association between plasma VEGF-A and progression-free survival and overall survival as researchers had observed in Avastin-treated metastatic breast, gastric, and pancreatic cancer patients.

[ pagebreak ]

In AVF2107g, metastatic colorectal cancer patients were randomized to receive irinotecan/5-FU/leucovorinIn regimen with or without Avastin. In AVAiL, NSCLC patients were randomized to receive gemcitabine and cisplatin plus or minus Avastin. In AVOREN, metastatic renal cell carcinoma patients were randomized to receive either Interferon-alpha2a plus Avastin or placebo until disease progression.

Similar to the other studies, researchers in this retrospective analysis categorized patients as falling into either low or high VEGF-A plasma levels based on pre-specified cutoff points, and analyzed whether patients with high VEGF-A levels experienced longer overall survival or improved progression-free survival compared to those with low levels of the biomarker.

Jayson et al. were able to confirm the prognostic value of plasma VEGF-A in all three disease settings. "Patients in the control group with high plasma VEGF-A levels had shorter overall survival than patients with low levels," the researchers reported. "However, potential predictive value for plasma VEGF-A was not seen."

While noting that previous studies demonstrated potential predictive value for plasma VEGF-A in breast cancer, pancreatic cancer and gastric cancer, "these findings were not replicated in metastatic colorectal cancer, NSCLC, and renal cell carcinoma," Jayson et al. wrote in the abstract, attributing the differences in findings to variations in sample handling in the different trials.

They added that the ELISA assay designed to gauge shorter VEGF-A isoforms, which was used in the breast, pancreatic, and gastric cancer studies, likely increased the test's sensitivity over previous assay models used in Avastin biomarker studies. As such, they hypothesized that the shorter isoforms — namely VEGF-A121 and VEGF-A110 — "are driving predictive value and might be diverse in different tumor types." Investigations to test this theory are currently ongoing.

Have topics you'd like to see covered in Pharmacogenomics Reporter? Contact the editor at tray [at] genomeweb [.] com.