The genomically driven therapy movement in oncology began in the late 1990s with the recognition of substantial clinical benefit from therapies like imatinib, tied to the presence of the bcr-abl fusion gene, ErbB2 and the use of trastuzumab in breast cancer, and BRAFV600 mutations in melanoma. In each of these cases, an activating mutation was associated with a therapeutic inhibitor, and the search for more targets was on.
Before we have all the answers on what treatments to give, we dove into the promise that sequencing would lead us there. When in doubt we send samples off to commercial or University owned labs in expectation of a result that will transform patient outcome.
So where are we in 2018? Maybe not as far as expected. In the JAMA article by Marquart et al, the authors explored the number of patients who underwent a genomic screening test and were subsequently considered ‘eligible’ to receive a genomically driven therapy.
Their methods are worth describing. They started by looking at the death statistics in the US for one year by type of cancer and back-calculated to the number of patients who might be eligible to receive a genomically guided test and therapy. For example, about 85% of patients who die of lung cancer have non-small cell lung cancer (NSCLC), and may benefit from genomic testing, while the 15% with small cell does not. Thus, they multiplied the denominator of 160,000 US deaths from lung cancer by 0.85 to arrive at the number of patients who should be tested because they could benefit from a genomically guided targeted therapy. From that amount, the number is further reduced as the incidence of activating mutations of the EGFR ( which would correspond to a choice of Afatanib, which is FDA approved for that indication) is only about 15%. Thus, the number of NSCLC cancer patients who should benefit from genomic testing is approximately 160,000 x 0.85 x0.15= 20,000 (approximately) Those are the ones who could benefit clinically, not necessarily those who did benefit. You get the idea. That’s 20% of the lung cancer patients, about 3% of the total of about 580,000 individuals who died of metastatic cancer in that year (2013).
But a key point to the study was the analysis they did of the trend over time. In 2006 only about 29,000 patients were eligible for genomically-guided therapy (because fewer therapies were available). That was about 5% of the total cancer patients. In 2018 they estimate the utility of genomic testing in 51,000 of the estimated 610,000 Americans who will die of metastatic cancers. That’s an increase from 5% to 8.3%. That’s only the patients who COULD benefit from genomic testing, not those who actually did.
The reason is that not everybody who takes a genomically-guided therapy based on a genomic test is going to benefit. The response rates vary but in many cases are well below 100%. Marquart and colleagues then finished their analysis by correcting for the factor of the response proportion for each of the targeted therapies. In the end, they conclude that only 4.9% of the total of US cancer patients benefit from genomic testing in 2018, but that this was increased from 0.70% in 2006 when fewer targeted drugs were available.
So what does this have to do with prostate cancer? Prostate cancer didn’t even make the list for a couple of reasons. We currently have really only two classes of therapies that are genomically driven and can be said to apply to prostate cancer: One is the PARP inhibitors, but they are not proven to work in prostate cancer yet despite a nice result from a study in the UK, not FDA approved yet and may not even require genomic testing to provide benefit (more to follow on this from ASCO 2018). The other is Pembrolizumab, which does require documentation of a hypermutated MSI high genotype – and I have covered elsewhere in my UroToday Blog: "Riding High on MSI"
So if, for the sake of discussion, we conclude that Pembrolizumab will benefit 85% of the 5% (optimistically) of CRPC patients who harbor an MSI high genotype, and PARP inhibitors will similarly benefit 85% of the 15-20% or so of patients who harbor a DNA repair defect, then the calculation is as follows:
I guess I’ll take those odds. Of course, I had to make some assumptions here, but if one in 5 of our mCRPC patients will benefit from the testing, that’s not a bad number to treat. By testing all mCRPC patients we will have to test 4 patients who will not benefit every 1 patient who will. And also, we have to give the PARP inhibitors the rigorous testing in randomized trials that will help us determine how good (or bad) my assumptions are.
But it’s a start.
Written by: Charles J. Ryan, MD