The jury in Elizabeth Holmes’ criminal trial heard from Theranos’ founder and former CEO for the first time yesterday. What she said, however, didn’t come from the witness stand—Holmes hasn’t been called to testify yet. Rather, the prosecution played interviews recorded by journalist Roger Parloff, who wrote a cover story about Holmes when he worked at Forbes magazine.

Over the course of 10 hours of taped interviews, Holmes made a number of remarkable, and damning, statements.

In one clip, Parloff asked her how many tests Theranos can perform. “I think can we say more than 1,000,” Holmes said.

In another excerpt, the jury heard Holmes telling the reporter that “the fact that we have a single device that can perform any test” was a trade secret.

The jury also heard Parloff asking Holmes whether Theranos could essentially replace Quest Diagnostics, one of the nation’s largest medical laboratories. Quest could do 600 some tests, Parloff pointed out to Holmes, including tissue biopsies.

“We can do all of those tests so we can provide data back to clinicians for all the same tests,” Holmes said after a brief hesitation. When asked why she hesitated and didn’t say “replaced,” she replied, “We’re processing the samples in different ways, let’s put it that way,” assuring Parloff that Theranos’ data is still of the “highest quality.”

Many of those statements were outright falsehoods or, more charitably, danced around the truth. In reality, Theranos couldn’t perform more than 12 tests on its Edison devices, former lab associate Erika Cheung told the jury earlier in the trial. Even then, “the Edison analyzer could only run one type of test for one patient at a given time,” Cheung said when she took the stand in September.

Holmes also shared with Parloff the doctored reports that sported various pharmaceutical company logos—including Pfizer, Schering-Plough, and GlaxoSmithKline. On the stand, Parloff confirmed that she never told him that the reports were produced by Theranos and hadn’t been approved by those companies.

Parloff also told the jury that Holmes never disclosed that most of their tests were performed using third-party devices, despite the fact that he asked her that question directly.

Article and correction

When Parloff’s article was published in June 2014, it became part of Holmes’ packet sent to woo investors. It’s easy to see why. In the article, Parloff described Theranos’ finger-prick method—“To me, it felt more like a tap than a puncture”—and he relayed the sheer number of tests the company could do at lower cost than competitors. 

“Theranos’s tests can be performed on just a few drops of blood, or about 1/100th to 1/1,000th of the amount that would ordinarily be required,” Parloff wrote, “an extraordinary potential boon to frequently tested hospital patients or cancer victims, the elderly, infants, children, the obese, those on anticoagulants, or simply anyone with an aversion to blood draws.”

For Theranos, the article was a marketing gold mine. The company even linked to it from its website.

Yet, in July 2015, at a Theranos demonstration at a board member’s law firm, Parloff noticed something odd. The company had set up two devices, one to run a potassium test, another to run an Ebola test. “I was a little surprised they needed two machines because I thought one could do everything,” he told the court.

Months later, after The Wall Street Journal published its investigation into Theranos, Parloff went back to Holmes to ask her to clarify points made in the article. Specifically, he asked her how many tests Theranos could perform on its proprietary device as of late 2014, the date mentioned in his article. “She responded, ‘50, 60, maybe 70. We can get you that number,’” Parloff told the court.

Ultimately, he released a “protracted correction” in which he picked apart several claims Holmes made to him.

Uneven test results

Earlier in the day, the jury heard from Dr. Mark Burnes and his patient, Merhl Ellsworth. Though Ellsworth’s time on the stand was brief, testimony from Burnes, who ordered blood tests for him with Theranos, was revealing.

In the summer of 2015, the time of the tests, Ellsworth was a retired dentist who was about to embark on a two-year mission trip to Asia. He had checked in with his doctor to ensure he was fit to travel. Burnes ordered a prostate cancer screening, which Ellsworth had undergone before, and he recommended Theranos since the company’s tests were cheaper than the competition.

What followed sounds strikingly similar to another patient’s experience: when the results came back, Ellsworth’s prostate-specific antigen numbers were high, at 26.1, a potential sign of prostate cancer. Four days later, Ellsworth took another test. This time, the results were 1.71, which were similar to earlier pre-Theranos results and did not indicate the presence of cancer.

Still, Burnes wasn’t confident in either number and called Theranos. The regional lab director told him the first result was probably an error. Burnes said he would have Ellsworth take the test for a third time, and he asked the regional director if Theranos would reimburse for the cost, which it did.

The third test came back high once more—22.9. Burnes called Theranos, telling the court that in the conversation, “I expressed a lack of confidence in their tests.” He asked Theranos to pay for yet another test, this time with a venous draw. The company agreed, and the result was normal—0.95.

Inside Theranos, executive Daniel Young and other employees had been discussing the botched results in a lengthy email chain, which the jury saw had been forwarded to Holmes.

In cross-examination, Holmes’ attorney asked Burnes whether he was familiar with the error rate for medical diagnostic labs. Earlier, Burnes had pointed out that in his 30 years practicing medicine, he has interpreted over 10,000 tests for prostate-specific antigen. He told the defense attorney that, yes, labs can have errors. But “they’re very rare.”

In March 2016, nearly a year after Theranos ran Ellsworth’s blood, the company sent Burnes a corrected lab report in which it had voided the high numbers. The PSA results, it said, ”should not be used out of an abundance of caution.”

https://arstechnica.com/?p=1814434