ASCO 2010: Chasing Waterfall Plots

ASCO Name Tag Logo Black SMost scientific conferences contain two very different approaches to storytelling. On the exhibit floor, pharmaceutical companies, laboratory suppliers, publishers and chambers of commerce compete for attendee’s attention with ever more grandiose booths that resemble small palaces. But in the many, many rooms surrounding that circus, scientific data is presented (usually) in decidedly unflashy fashion, thanks to the understated personalities of most scientists and a sense of research presentation propriety. 

The American Society of Clinical Oncology meeting illustrated that divide more vividly than any other scientific conference I’ve attended. Just last month, I was impressed by the exhibit booth architecture display at the BIO meeting, but the exhibitor showcase at ASCO made other conferences’ huts look like a miniature golf course. With Jumbotron-sized video monitors, 3D animations, futuristic chairs, and more sales reps per square foot than anywhere else in the world, it’s an overstimulating experience to walk through.

But the research presentations of ASCO are the polar opposite – quick-fire 15-minute “just the data” talks light on drama even as exciting new treatments are described. The structure for presenting clinical trial results is apparently very codified for cancer researcher: provide rationale for trial, describe study design, show effect of treatment (or lack thereof), compare side effects to other treatments, suggest future directions. Context for the data is later provided by a discussant, who is at more liberty to deliver a judgement on whether the findings have immediate clinical relevance. Whether that’s down to ethics or merely tradition, it’s a much more formal experience than other conferences I’ve seen.

The overwhelming size of the conference makes it hard to deliver a 30,000-foot overview of the proceedings; one must merely pick a particular cancer track and hope for the best. On Monday, following the University of Chicago trail led me to a series of presentations on head and neck cancer, a particular specialty of the Medical Center and our current CEO, Everett Vokes. Like all cancer types, the head and neck variety can be broken down into several sub-types, including thyroid cancer, which can be further broken down into its own sub-types. One disease on the smaller end of this Russian nesting doll hierarchy is medullary thyroid cancer (MTC), diagnosed in only 1,400 Americans a year but essentially untreatable.

Because of its rarity, MTC is also very difficult to study – to test a new drug treatment thoroughly, a trial needs hundreds of patients. But three trials at this year’s ASCO meeting scraped together patient pools large enough to find encouraging results from drugs already approved for use in other cancers. Vandetanib, sunitinib, and the still unnamed XL 184 all work by similar mechanisms (blocking the growth factor receptor VEGFR and the tyrosine kinase RET) and all three showed promise in shrinking tumors and interrupting progression of the disease. 


A sample waterfall plot.

Those presentations also deployed a graphical representation of success that was new to me – not the Kaplan-Meier curve central to Malcom Gladwell’s New Yorker story, but a “waterfall plot.” Appropriate for smaller studies, the waterfall plot charts each individual patient’s response to the drug on a parameter such as tumor size. Unlike an average of patient responses, or a response rate without information about the magnitude of response, the waterfall plots presented by Jonas De Souza of the Medical Center and Razelle Kurzrock of MD Anderson Cancer Center strikingly demonstrated the variability of an experimental treatment. While tumors from a few patients were unperturbed by the drug and continued to grow, the majority of patients in both studies showed at least some decrease in tumor size – good news for a disease that had previously frustrated oncologists.

For other types of head and neck cancer that already have treatments, researchers endeavored to improve the current standard of care. The Medical Center’s Tanguy Seiwert presented data comparing a new drug, BIBW 2992, to cetuximab, a popular therapy for squamous cell carcinoma of the head and neck despite only a 13 percent response rate and a high incidence of drug resistance in patients. In a small Phase II trial, BIBW 2992 nearly doubled the response rate of cetuximab, but also showed much higher rates of one particular side effect: diarrhea. Data presented in an evening session described the use of several biomarkers to better predict the severity of a patient’s cancer, tools that could be used to adjust treatments to avoid such negative side effects. But as the University of Chicago’s Ezra Cohen commented as the session’s discussant, the time to validate such biomarkers is very long, and can only be accomplished through the rigorous collection of specimens from large clinical trials.

Like other clinical trial conferences I’ve covered recently, such a discussion hinted at a sea change in the field – away from the traditional drug group vs. control group and toward more personalized treatments. In that light, waterfall plots felt like the more appropriate shorthand: figuring out why a drug worked in the highly-responsive patients on the right and not the progressing patients on the left will be the central challenge for oncologists in the 21st century. Perhaps to address these technically complicated questions of personalized medicine, the next generation of researchers can borrow some technology from their neighbors on the exhibit floor.

About Rob Mitchum (525 Articles)
Rob Mitchum is communications manager at the Computation Institute, a joint initiative between The University of Chicago and Argonne National Laboratory.
%d bloggers like this: