The Eighty-Million-Dollar Miss: A Year Inside the Zenvara Launch

The chart below, reconstructed from the company’s IQVIA prescription panel, is the story in a single image. The plan line is what the company committed to in February 2025. The actual line is what arrived, week by week, after the launch in early May.

The two lines run together for the first thirteen weeks. From there, they separate in three phases. The first separation, beginning in week fourteen, is small and explicable. The second, beginning in week twenty-seven, looks like partial recovery. The third, beginning around week forty, is the one nobody saw coming, because by then the company was watching for a different kind of failure entirely.

What follows is each phase, in the order it happened.

The drug shipped to specialty pharmacies on May 5, 2025. Through the first thirteen weeks, new-to-brand prescriptions tracked within four percent of the forecast in every region of the country. The field force logged just over six thousand calls in the first quarter, a fifth of them lunches with oncology practice managers, almost all of them face-to-face. The Veeva CRM records from this period, read months later, contain a single repeated phrase from representatives across territories: good clinical reception, no access blockers.

By the end of July, the company was at ninety-eight percent of forecast for cumulative prescriptions and the launch lead was, on paper, ahead of schedule. The first internal post-launch review, held in August, recorded no concerns. The commercial team began discussions about pulling forward investment in a second-line indication.

What the company did not yet know was that two of the three largest commercial insurers in the country, watching the same IQVIA file from the other side of the table, had already begun reviewing the prior-authorisation criteria.

The first formulary change arrived quietly. On August 18, Aetna’s 2026 formulary draft, circulated to manufacturers under a non-disclosure agreement, moved Zenvara to tier three with a new step-therapy requirement through generic carboplatin and pemetrexed. The change would not take effect until January, but the signal it carried — that the country’s third-largest commercial insurer was preparing to throttle access — was actionable in August.

It was acted upon. The market-access director, Aisha Bello, was four weeks into a six-week parental leave; the Aetna draft was caught the next morning by Vikram Iyer, the market-access manager named on Dana Park’s coverage memo. He posted it to #market-access-payers at 8:45 a.m. on August 19 and asked the team for a read on the step-edit comparator. Mr. Reyes replied thirty-seven minutes later. The comparator was the cheap one. The team had modeled the step-edit before. The historical rate at which carbo/pem steers new starts away from the branded option was twenty-five to thirty percent, with about a ten-percent prior-authorisation-failure rate on the back end. By the close of business on August 20, Mr. Iyer had updated the Win/Loss Log with the Aetna outcome classified as B+: manageable, four million dollars of modeled Q1 revenue impact.

When Ms. Bello returned in the first week of October, she read through the backlog with Mr. Iyer and concurred with the classification. The IQVIA data by then was four percent under forecast — small enough to attribute to seasonal noise. Mr. Reyes’s weekly tracker entries from October refer to the gap twice and dismiss it both times. “Slight softening Q3, expect Q4 recovery as Cigna ramps.”

Two other things happened in the same window that the team would later wish it had connected. In mid-September, the specialty pharmacy partner serving Memorial Sloan Kettering reported its Zenvara turnaround had slipped from two-and-a-half days to five or six. The pharmacy attributed it to its own CoverMyMeds-to-Surescripts e-prior-authorisation migration, a known and dated transition. Mr. Reyes logged the explanation in #launch-war-room and did not escalate. A week later, the Food and Drug Administration amended the label for AstraZeneca’s Tagrisso to include the adjuvant resectable non-small-cell lung cancer indication based on the FLAURA-2 trial. The competitive-intelligence analyst, Jordan Patel, filed a one-pager framing the impact as second-line risk: patients relapsing from adjuvant osimertinib who might, eventually, reach Zenvara. He did not model the more immediate effect, which was that large oncology groups began re-sequencing their reflex biomarker testing to lead with EGFR rather than PD-L1. Patients who would have started on Zenvara six months earlier were now starting on osimertinib instead.

In early November, Cigna confirmed what the company had been negotiating since August: Zenvara would be added to its preferred commercial formulary effective January 1, with prior authorisation only. The win added roughly fourteen million commercial lives. In the moment, it read as the counterweight to the Aetna tightening — though it was not, in any honest accounting of timing. Formulary wins translate to scripts on a fourteen-to-eighteen-month lag, and the Cigna ramp would not produce meaningful Q1 volume in 2026. The Aetna step-edit would.

The chart, accordingly, looked stable for a quarter. From weeks twenty-seven to thirty-nine, the gap to forecast hovered between ten and twelve percent — uncomfortable, but consistent with what the team now believed was a one-time access correction that had been resolved.

On January 20, 2026 — three weeks after the Aetna change took effect — the first post-effective MMIT data arrived. The step-edit was routing sixty-one percent of new starts through the carbo/pem comparator, not the twenty-five-to-thirty percent the August model had assumed. The reason was visible in the same digest, if anyone had thought to read it that way: Aetna had quietly paired the formulary change with a tightening of its prior-authorisation criteria. The new criteria narrowed the eligible PD-L1 expression threshold and required pre-treatment imaging within sixty days. The two effects compounded rather than stacking. Mr. Iyer reposted the original August thread the following morning. Mr. Reyes revised the Q1 forecast down by eleven million dollars two days later. Nobody flagged the revision externally; the chief commercial officer learned about it in the year-end review.

The decisive turn came not from a competitor or a formulary committee but from the operational machinery of the country’s largest insurer. Between February and April, UnitedHealthcare’s commercial prior-authorisation pipeline for oncology drugs in the Northeast slowed. The cause, when it was identified by a consulting firm engaged in May, was a vendor transition: UnitedHealthcare had migrated PA processing to a new utilisation-management contractor, and the new workflow had not yet been calibrated for oncology turnaround times. Average decision time slipped from four days to nine.

None of this was announced. The MMIT digest recorded a tightening of the formal PA criteria on April 22, and Ms. Bello read it the following Tuesday in her regular Wednesday review. The five-day operational delay in turnaround, however, is not the kind of thing MMIT records. It would appear in the data only as unfilled prescriptions and abandoned starts, two to four weeks downstream.

By the time the IQVIA file for the week of May 8 arrived — the now-famous WK19 file that ended the year — the trailing four-week variance had grown to sixteen percent and was widening. National total prescriptions were 763 against a plan of 944. New-to-brand prescriptions in the Northeast were 354 against a plan of 412. The file landed at five minutes past eight on Monday, May 11. It sat on a shared drive for four days.

When Mr. Reyes opened it on Friday afternoon, the year was effectively over. The Cigna ramp would not catch up until the following fiscal year. The Aetna step-edit, four months in, had cost an estimated twenty-two million dollars against the original Q1 forecast. The Tagrisso share-shift, mostly invisible in the company’s own systems, had reduced the addressable 1L mixed-biomarker new-start pool by an estimated eight percent. The UHC operational delay was, at that point, costing roughly one million dollars per week in deferred or abandoned starts.

The first message that connected the year’s evidence into a single statement appeared on a Slack channel called #payer-escalations at 9:15 a.m. on Friday, May 15. It was three months too late to change the outcome and three weeks too late to prevent the worst of the Q4 slip. It is, by the company’s own admission, the moment everyone had finally seen the same picture.

Ms. Bello had the formulary change in her digest since April. Mr. Ortiz had been hearing about PA delays in his district meetings since March. Mr. Reyes had the IQVIA file showing the volume miss for the previous five days, and the four-week trend showing it for the previous month. The cost of the thread not happening earlier — in February, when the first whispers came from the field, or in January, when the Aetna step-edit revealed itself to be twice as severe as modeled, or in October, when the IQVIA gap first opened — was, by the most conservative internal estimate, sixty-four million dollars in foregone first-year revenue. The remaining sixteen million dollars of the eighty-million-dollar miss were attributed to the Tagrisso share-shift, which the company had not modeled at all.

On the Monday following the WK19 file, Mr. Reyes updated the weekly launch tracker — the Google document that records, in narrative form, the week’s commercial performance for circulation to the executive team. The relevant entry, the one cited in the chief commercial officer’s year-end review, reads as follows.

The numbers in the tracker are the numbers in the IQVIA file. The cause is the UHC PA latency. The action item is the payer call. There is no record, in this document or any other, of the year-long pattern that culminated in WK19. There is no document that ties the August formulary signal to the October gap to the February field complaints to the April MMIT digest to the May volume miss. The five documents that record those events live in five places, on five cadences, owned by five people. The synthesis was, in the company’s structure, no one’s job.

The diagnosis, on the evidence of a year, is not that any single system failed. Each system did what it was built to do. IQVIA shipped the file on time, every Monday, for fifty-two weeks. MMIT logged every formulary change. Veeva accepted the call records. The competitor abstract was uploaded to the American Society of Clinical Oncology submission system on schedule. The doctors saw the patients and the patients went home with whatever they could fill.

What failed was the synthesis. Five systems, each operated by a different person on a different cadence, held five fragments of one year. Mr. Reyes had the volume trend, read weekly, sometimes biweekly. Ms. Bello had the formulary picture, updated monthly. Mr. Ortiz had the field signal, surfaced anecdotally at district meetings. Mr. Patel had the competitive landscape, assembled annually. The HEOR director, in a different jurisdiction, had a separate twelve-month evidence horizon for the European submission of the same molecule under a different brand name. Each of the five people knew everything they were supposed to know. None of them knew enough to act in time.

The cost, in the industry, is well-documented. Between fifty-six and sixty percent of drug launches miss their first-year revenue expectations. ^[†]Sedulo Group, ‘Why 56% of Drug Launches Miss Expectations,’ 2025; MamaHealth, ‘Why 60% of Commercial Drug Launches Fail,’ 2026. Limited market access is the most-cited cause. Inadequate understanding of market and customer needs is the second. ^[†]Pharmaceutical Commerce, ‘Merging data for managing pharma commercialization.’ Neither cause is, strictly, a failure of clinical knowledge or strategic judgement. They are failures of attention, distributed across a workforce that was, in every individual instance, paying perfectly good attention to what was in front of them.

The eighty-million-dollar miss did not stay inside the commercial organisation. The Q3 2026 marketing budget was reduced by thirty-five percent at the board’s direction. The launch-incentive bonus pool was reduced by eleven million dollars; field representatives who had hit their individual quotas were paid out at sixty-two cents on the dollar. Two of the top five Northeast representatives left for competitors before September. The market-access organisation was restructured: the digest-review cadence moved from weekly to daily, and a second tier of payer-account ownership was added so that no single inbox carried more than four named payers.

The chief commercial officer who had set the original forecast resigned in July. Her successor commissioned an external review of the year’s data architecture, the report from which is, in part, the underlying source for this article. Mr. Reyes was retained and promoted in November to head a new cross-functional ‘launch-signals’ group with explicit responsibility for the kind of synthesis that, in the year described above, had been no one’s job. Mr. Iyer was made market-access director when Ms. Bello moved to a regional president role at a larger competitor in October. The Aetna step-edit was, by year two, modelled at the firing rate it actually fires at.

For the record, the timeline below tracks what was known, in which system, by which person, on which day across the twelve months of Zenvara’s first year on the market. The events are ones that, with hindsight, could have been assembled into a single picture by any observer with simultaneous access to all five files. No one had that access. By the time someone did, in a Slack thread on May 15, 2026, the forecast was unrecoverable by roughly eighty million dollars.