Mrs. M was a 46 year-old woman who, despite having never smoked, was diagnosed with metastatic adenocarcinoma of the lung. She had a nice initial response to chemotherapy, and when she eventually progressed, she was treated with Iressa (an EGFR inhibitor similar to Tarceva which is no longer available in the US). To both her and her doctor’s delight, she had near resolution of her lung mass and most of her liver lesions after 2 months on Iressa. Unfortunately, after remaining stable for 10 months, her restaging CT scan showed one lesion in the liver was growing. Despite holding the course for another 2 months, it became evident that Mrs. M was progressing in the liver, even though her lung mass remained as stable as ever it had over the preceding year.

If you’ve ever wondered why a patient with lung cancer can have a beautiful response in one area while growing in another, or why new metastases can develop while everywhere else the disease remains stable, you aren’t alone. The truth is that cancer heterogeneity (different cancer cells or tumors in one patient behaving differently from one another) has been observed for a long time. Studies of induction chemotherapy in lung cancer (chemotherapy given in early stage disease prior to surgery) performed in the 1990’s routinely reported response rates in excess of 60%, while we know that the same chemotherapy in metastatic disease has a response rate of only 25-30%. Clearly there is something fundamentally different in the metastatic cancer than in the early stage cancers. So what gives? This concept gets to the center of the question of why cancer can initially respond to chemotherapy but eventually becomes resistant.

We know that the normal genetic machinery of lung bronchial cells can be altered by exposure to carcinogens (like tobacco smoke) and, with some contribution from underlying inherited predisposition, can give rise to lung cancer. These cancer cells frequently lose the ability to repair DNA damage, so mutations accumulate as they divide. There is an extensive body of research into the mechanisms of how tumors develop the ability to spread (metastasize) throughout the body. What is also becoming clear is that during the development of the billions of cancer cells in a patient (1 billion cells = a 1 cm tumor) the cells can develop widely different genetic makeup and behaviors, making treatment a real problem.

To illustrate this point I’ll use a study from last year by Jeff Engelman (abstract here), a lung cancer scientist at MGH in Boston. He took tumor samples from patients with lung cancer, all of which had mutations in the epidermal growth factor receptor (EGFR) that predisposed them to responding to Iressa. Eventually they progressed, and he was able to analyze changes in the EGFR gene from when they started (and responded) to when they progressed. What he found was that about half of them had developed a new mutation (called T790M) that made them resistant to Iressa. In about 20% of the resistant tumors, though, there was a totally different mechanism of resistance. The tumor cells had developed the ability to make more of a protein called MET, which could work with the EGFR to overcome Iressa. The most interesting thing from this paper was an observation in one patient (red box below) where progression was seen in both the lung and a mediastinal lymph node. He found that the primary tumor had developed the T790M, while the node did not have T790M but instead had MET amplification, leading to (at least) 2 different mechanisms of resistance in one patient.

(Click to enlarge)

This has important ramifications for the future treatment of lung cancer. Even if we are eventually able to perform routine biopsies and determine a marker of benefit from a treatment (like EGFR mutations and Iressa/Tarceva), we now have to assume that some of those billions of cancer cells have already developed one or perhaps many ways around that treatment. Our current genetic assays are usually not sensitive enough to pick out a single mutation from the millions of other “normal” cancer cells, but even if we could we would not normally biopsy every metastatic lesion in a patient to verify that all of them had the same marker. We cannot even be sure that all of the progressing lesions in a patient are progressing for the same reason!

One way around this would be to get a more global picture of the cancer through blood tests of circulating cancer DNA, proteins, or even circulating tumor cells (CTCs), which might reflect all of the genetic changes present in the patient. Shyamala Maheswaran published a study in 2008 (described in prior post here) showing that CTCs could be used to detect very low levels of the T790M mutation in lung cancer patients with EGFR mutations, and that this predicted that the patients would eventually progress. As new mechanisms become know, this technology could be used to monitor patients for emerging resistance and allow us to adapt new drugs to overcome it. Another possible way around this is simply to begin treatment with a combination of drugs that overcome potential mechanisms of resistance (see below).

(Rx is a medical abbreviation for "treat")

Although today it seems like every time we whack the mole/cancer that we see on the head with our hammer/treatment, another pops up right next to it, we are clearly gaining an understanding of why we have been failing up to this point. The era of simply using a single “best” treatment for first, second, or nth-line treatment is nearing its end, hopefully to be replaced by more rational treatments that take into account individual tumor vulnerabilities. At the risk of carrying my analogy too far, I hope in my lifetime we are able to anticipate where that mole will pop up and be waiting there, hammer in hand.