Oh, Canada. “Limits to Personalized Cancer Care” in the NEJM

The NEJM, despite its habit of publishing every precision medicine milestone in the past 15 years (including the Herceptin and Zelboraf papers cited in the first paragraph of this editorial), elected to give voice to the naysayers: per an opinion from physicians at the Princess Margaret Cancer Centre, treating cancer based on genomic profiling is unproven, expensive, toxic, and requires further investigation in controlled trials before we should consider it as a treatment strategy. No shock that this came out of Canada, that hotbed of medical innovation.

The authors do make some claims that are fundamentally true. I have a physician friend who likes to say, “It’s cheaper to let people die.” This is a fact and cannot be disputed. Similarly, cancer therapies are associated with toxicity. Combining therapies may increase toxicity. It’s a shame that there are no explicit instructions for reducing the dose to limit these toxicities, right in the label or something.

It may be worthwhile, however, to focus this rebuttal on information that is incorrect, outdated, or deliberately misleading.

First, the authors hold targeted therapy to a ridiculous standard. “… For most patients with metastatic cancer, the duration of benefit [from targeted therapies] is limited and is followed by drug resistance and cancer progression.” Remind me of the cure rate for chemotherapy in the metastatic setting, please?

Let’s judge targeted therapy by the same standard as any anti-cancer therapy: progression-free survival (PFS). The editorial has a table that depicts the many failures of targeted therapies:


The first row is dedicated to the damning SHIVA trial, which concluded in 2015  that “The use of molecularly targeted agents outside their indications does not improve progression-free survival compared with treatment at physician’s choice” and “Off-label use of molecularly targeted agents should be discouraged.”  However, the study employed a crossover design that allowed patients randomized to physician’s choice to crossover to a targeted therapy upon progression. A subgroup analysis of these patients presented in a poster at ASCO in 2016 compared PFS for the targeted therapy with the physician’s choice therapy, essentially using the patients as their own controls. In 35% of these patients, PFS on the targeted therapy was 1.3x the same subject’s PFS on physician’s choice therapy. It is worth noting that PFS generally decreases with each line of therapy.

The MOSCATO trial enrolled 1,100 patients with a variety of cancer types, performed genomic sequencing on a fresh biopsy, and assigned patients to a targeted therapy based on the results. Each subject’s PFS on prior therapy was used as the control PFS. Of the enrolled patients, 49% had an actionable target, and 19% were treated with a matched therapy. The authors reported that 33% of patients treated had at least a 30% increase in their PFS with the targeted therapy, and 62% of patients experienced disease control (the combined rate of subjects with a response or stable disease). Further, one of the authors is quoted as follows: “[MOSCATO] specifically excluded patients with well-established actionable targets for which approved and marketed targeted drugs are available (i.e., lung cancer with EGFR mutation or ALK translocation, B-RAF mutant melanoma, GIST with KIT mutations or breast cancer with HER2 amplification.)” This is remarkable, because the design demonstrates that targeted therapies work outside their approved indications and that genomic targets can be pan-cancer (another point disputed in the editorial). And while the press release linked is dated September 23rd, entirely too late to be represented in the NEJM commentary, the study was not included in the table of supporting evidence as one of the results-pending entries.

As mentioned above, the authors argue against using a targeted therapy outside its approved indication. This sentence caught my eye: “Vemurafenib (Zelboraf) was associated with a higher rate of survival than dacarbazine among patients with melanoma that expresses the BRAF V600E mutation but had only modest activity against other biomarker-selected cancers that express this mutation sporadically.” This … did not sound true. And it’s not. Published in the NEJM in 2015, the paper cited concluded that “In the cohort with non-small-cell lung cancer, the response rate was 42% (95% confidence interval [CI], 20 to 67) and median progression-free survival was 7.3 months (95% CI, 3.5 to 10.8).”

Well, then. But what of the cohort with Erdheim-Chester disease or Langerhans’-cell histiocytosis, you ask? The response rate was 43% (95% CI, 18 to 71), with a median treatment duration was 5.9 months (range, 0.6 to 18.6).

So they did great, too. Not to be overlooked: “There were anecdotal responses among patients with pleomorphic xanthoastrocytoma, anaplastic thyroid cancer, cholangiocarcinoma, salivary-duct cancer, ovarian cancer, and clear-cell sarcoma and among patients with colorectal cancer who received vemurafenib and cetuximab.”  Oh, and, “In three of these patients (one each with anaplastic thyroid cancer, cholangiocarcinoma, and ovarian cancer), the responses have persisted for more than 12 months.” A patient with advanced ovarian cancer responded to monotherapy for 12+ months. No big deal. That happens all the time.

“Modest” activity, indeed.

The editorial also takes issue with the lack of centralized trials, the waste and exhausted resources inherent in a system where everyone is running their own sequencing studies. There’s no reason to spend a lot of time on this one, because U.S. healthcare is decentralized, and many of the major U.S. cancer centers deployed their own internal sequencing panels years ago. This makes a lot more financial sense in large-volume centers than sending the samples out and letting the sequencing company bill for the analysis. MDA has one of these in-house panels, and they just happen to be providing sequencing services for the large, biopsy-challenged, federally-funded MATCH trial (where, as of May 2016, 2.5% of 795 screened patients managed to get on treatment). Note from the MATCH interim analysis that patients died waiting for their biopsy results and treatment assignments. MATCH’s defense? The lab got backed up. In a system where I have a choice to enroll in MATCH or to send tissue to one of the multiple private companies who will perform the sequencing and have the results back to me in two weeks, why be a martyr?

Individual claims aside, this editorial represents the backwards-thinking medicine that threatens patients. It implies that the system has gone rogue, advocating blindly for unproven, dangerous therapies, as if the gold standard were so effective and we’re all falsely enchanted by novelty alone. Chemo doesn’t work. What choice do we have? Physicians don’t want to treat with chemo anymore. There are two or three chemotherapy agents that I haven’t tried that I could reasonably attempt, but I’ve been at my current center for a year and a half, and my oncologist hasn’t offered them. I haven’t asked. But I’ve also seen evidence of how married major centers can be to the SOC. I was at a conference and heard a physician say, onstage, at a targeted therapy session, “Chemo cures people. This stuff never cured anyone.”

I’m waiting for results from a sequencing test. Despite the Canadians’ predictable condescension – “There should also be a clear message to patients that personalized cancer medicine has not led to gains in survival or its quality and is an appropriate strategy only within well-designed clinical trials” – I’ll go where it leads me.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s