We know that cancer cells mutate over time -- in fact, that's how they became cancer cells. In fact, oncologists see the heterogeneity of cancer cells in our daily practice every day. Although there are certainly many reasons why patients have some of their cancer cells respond and others not (blood supply, local microenvironmental factors, etc.), one of the important factors is genetic heterogeneity within the cancer cell population -- some cells die and others don't. Moreover, the typical gradual shift from a cancer responding to a treatment and then showing an acquired resistance and progression after several cycles is presumably mediated by an ongoing evolution of the tumor itself to selective survival of cancer cells with a new resistance mechanism.

I suppose it should come as no surprise, then, that EGFR mutations are also subject to tissue heterogeneity. This is demonstrated by a paper in this month's Annals of Oncology by Gow and colleagues, out of Taiwan.This paper checked for EGFR mutations in tumor tissue from 67 patients who underwent some form of surgery of biopsy or resection of NSCLC in Taiwan that included removal of some tissue from both the primary tumor and a metastatic lesion. Those 67 paired patient samples came from a pool of over 900 surgical samples over an 8 year period; none had received an EGFR inhibitor at the time that tissue was collected.

They found that 52% of the patients had a mutation (underscoring how remarkably different the biology of NSCLC is in Asia vs. North America, where we'd expect about 10%). Of the 18 patients with an EGFR mutation in the primary tumor as detected by DNA sequencing, 9 (50%) didn't have a mutation detected in the paired metastatic lesion. Of the 26 patients who had an EGFR mutation detected in the metastasis, 17 (65%!) didn't have one in the primary tumor.

Finally, they compared EGFR mutation testing using two different techniques, DNA sequencing and one called the Scorpion Amplified Refractory Mutation System, noting a 27% discrepancy rate in detection of EGFR mutations between the two methods.

I don't think these results are suprising, but they're humbling. We've seen that even in a patient population with an EGFR mutation, the response rate to EGFR TKIs sn't 90-100% but about 70-75%, while some people without a detected EGFR mutation can have a great response. We also see that patients who respond to EGFR TKIs often have areas of residual disease, or a mixed response with great tumor shrinkage in the lung but progression in the bone. These observations likely correspond to the differing tumor biology of the different areas of tumor.

Of course, it's also humbling to know that we can go to the effort to biopsy a tumor just to send off for an EGFR mutation and still get a non-representative result, depending on the area of tumor sampled and even the technique used.

As a growing consensus emerges around the importance of EGFR mutations, it looks like these issues will remain a major barrier. The fact that only a minority of patients with advanced NSCLC have tissue available for sending for biomarker correlates is already a major challenge. Knowing that multiple biopsies from different tumor areas may to actually be needed to really characterize the EGFR status of a patient's cancer is a huge problem.

This is still very early work, but it's an important wrinkle in the emerging EGFR story.