The Office of Prescription Drug Promotion put Genentech on notice with an Oct. 3 untitled letter saying the company’s visual aids for its cancer drug Tarceva were misleading, and ordered the drug maker to pull the pancreatic cancer and NSCLC visual aids.

Among the agency’s objections: that the pancreatic cancer visual aid “drastically overstates the efficacy of Tarceva,” by saying using the medicine with gemcitabine boosts overall survival by 3.7 months. The problem: the FDA-approved patient information pegs the survival benefit as 12 days, or 6.4 months with Tarceva/gemcitabine as opposed to 6 months without Tarceva in the mix.

This same visual aid ran afoul of the agency because it tied this same misstated survival benefit to a type-2 rash that some patients may develop and portrayed the rash as a sign that the drug is working. The government watchdog wrote this is a problem for two reasons: first, the rash and time frame weren’t part of the original study’s endpoints, and second, because it portrays a rash—an adverse reaction—“as an efficacy predictor and therefore a potential benefit to patients.” The agency also pointed out that the PI warns patients not to tweak their drug regimens to trigger rashes because “data do not support increasing the dosage of Tarceva to cause patients to develop rash.”

OPDP also took issue with wording in the Tarceva NSCLC visual aid, saying the language in question suggested clinical significance that it couldn’t back up. In this case, the agency wrote the company used retrospective analysis to back its visual aid claims that it reduced the risk of death among squamous cell lung cancer patients by 33% vs placebo, and by 29% among adenocarcinoma patients on Tarceva vs placebo. OPDP wrote that although Genentech collected information about tumor histology as part of the trials, spinning that information after the fact is a problem because it’s not supported by “adequate prospective statistical design,” meaning the numbers lack the heft to be considered substantial evidence or substantial clinical experience.

The critique wasn’t limited to math—the regulatory group also dinged Genentech for imagery it claims oversells just how long Tarceva could keep patients alive. OPDP wrote that the company’s use of an hourglass showing a grandparent and child was a problem because the sand doesn’t  flow, suggesting “that time is standing still for cancer patients because of Tarceva therapy.” The agency pointed out this is a bit of a reach because the pancreatic cancer benefit hovers around 12 days of additional survival, and one month for lung cancer patients. “These improvements . . . do not support the implication that Tarceva can slow disease progression and greatly improve survival,” wrote OPDP.

Update: A Genentech spokesperson emailed MM&M that the company is addressing the FDA’s concerns and is making appropriate changes.