Drug discovery tools · 3D tissue models
zPREDICTA
A San Jose company that rebuilt human tissue in a well plate so cancer drugs could be tested on something closer to a patient. Y Combinator backed it in 2015, LabCorp partnered with it in 2020, a Nasdaq buyer paid about $10 million for it in 2021, and thirteen months later the buyer carried it at zero.
A clinical trial in a test tube
A technician mixes cells into a viscous liquid and pipettes it into a well. The liquid is not a generic gel. It is a formulation of collagens, laminins, fibronectin and hyaluronic acid blended to match the extracellular matrix of one specific human organ, and the company said its matrices had more than 90 percent identity with human tissue. At room temperature it polymerizes. Growth medium carrying the cytokine environment of one specific disease goes on top.
Over the next few days the cells build themselves into spheroids, tubes and networks. The layer runs about 800 micrometres to a millimetre thick, and primary human cells stay alive in it for at least 21 days.
That was the product. zPREDICTA called the platform PD-3D, for Patient-Derived 3D Tissue Culture, and sold reconstructed organs under names that read like a parts catalogue: r-Bone for multiple myeloma and acute myeloid leukemia, then r-Lung, r-Breast and r-Stomach. In its Y Combinator application the company repeated what a customer had called it: “a clinical trial in a test tube.”
The problem it was aimed at
Most cancer drugs fail. The FDA, writing in April 2026, put it bluntly: historically more than 90 percent of drugs that clear animal studies do not get approved. zPREDICTA’s argument was that the failure starts earlier than anyone admits. Cells grown flat on plastic have no architecture and no neighbours. Mouse models have both, but they are mouse. Neither reproduces the matrix, the stroma and the soluble factors that make a human tumour shrug off a drug that looked potent in a dish.
The regulatory ground under that argument moved twice, both times after the company had been sold. The FDA Modernization Act 2.0 was signed into law on December 29, 2022, ending the statutory requirement for animal testing of new drugs. In April 2025 the FDA published a roadmap to reduce animal testing in preclinical safety studies and said it would promote lab-grown human organoids and organ-on-a-chip systems.
Two scientists who left the university
Julia Kirshner took her PhD at City of Hope in 2003, trained at Lawrence Berkeley National Lab and the University of Alberta, and joined the faculty of Purdue’s Department of Biological Sciences in 2008. The laboratory work is in the record: a 2008 paper in Blood describing a three-dimensional model for evaluating the impact of therapy on multiple myeloma, an NIH R21 of $158,399 in fiscal 2011, and a 2014 protocol paper in the Journal of Visualized Experiments written with her eventual co-founder, Mukti Parikh, who became chief technology officer. The two met at Purdue. Kirshner left academia in 2014.
They took an unusual route for a wet-lab company, going to Y Combinator rather than a life science incubator and appearing at the Summer 2015 demo day. Kirshner was direct about why: the science was covered, the business experience was not.
Customers first, capital second
The commercial history is thin on rounds and thick on partners. Aggregator data records about $220,000 of private funding across two rounds, the second a 2016 convertible note raised through Propel(x). No Form D was filed under the company’s own name, and no federal grant was awarded to it.
What it had instead were buyers of services. An ImmunoGen antibody drug conjugate was tested in the model in a 2016 paper co-authored by ImmunoGen scientists and Kirshner. In November 2018 the models went onto the Science Exchange marketplace. In February 2020 LabCorp signed a strategic collaboration to develop and commercialise tumour-specific preclinical models on the platform, tying zPREDICTA to Covance. Terms were not disclosed. Headcount stayed tiny, recorded at three people at the end of 2022.
Thirteen months
On November 24, 2021 Predictive Oncology Inc, a Nasdaq-listed company then based in Eagan, Minnesota, acquired zPREDICTA by merger for $10,015,941 in cash. Kirshner signed an employment agreement and became president of the zPREDICTA division. The buyer told investors the deal would be accretive in the short term as well as the long term.
By June 30, 2022, seven months later, the buyer had tested the zPREDICTA reporting unit with a discounted cash flow model using a 65 percent discount rate and concluded the goodwill was worthless. It wrote off $7,231,093. In the same quarter the unit booked about $0.1 million of sales, and Kirshner was promoted to chief scientific officer of the parent. In the fourth quarter the remaining $3,349,375 of acquired intangibles went to zero too. At the end of 2022 zPREDICTA Inc was merged into Predictive Oncology and folded into the Pittsburgh segment. In September 2025 the buyer adopted a digital asset treasury strategy and closed about $343.5 million of private placements to fund it.
What is proven, and what is still claimed
| Evidence | What the record shows | Source type |
|---|---|---|
| Acquisition | Merger closed Nov 24, 2021 for $10,015,941 cash. The founder became president of the acquired division. | Public record |
| What was bought | Intangibles $3,780,000: developed technology $3,500,000, customer relationships $200,000, trade name $80,000. Goodwill $6,857,790, later adjusted to $7,231,093. | Public record |
| Revenue | Unaudited pro forma combined 2021 revenue $2,429,786 against the buyer’s own $1,420,680. zPREDICTA booked $352,379 in the year ended Dec 31, 2022. | Public record |
| Written to zero | Goodwill fully impaired at June 30, 2022 using a 65 percent discount rate. Intangibles of $3,349,375 impaired in the fourth quarter of 2022. | Public record |
| Accretion promised | The December 2021 release called the deal “accretive on a short term as well as a long term basis”. The filings show the whole price impaired within thirteen months. | Claim differs from the record |
| Literature | Kirshner’s 2008 Blood paper, a 2014 protocol in the Journal of Visualized Experiments, a 2016 ImmunoGen study of IMGN901 in the model, and a 2016 ChemMedChem paper carrying a zPredicta affiliation. | Public record |
| Patents | Application US20190169562A1, assigned to zPredicta, Inc., status abandoned. Aggregator records show three filings and one grant dated Sep 21, 2021. | Public record |
| Clinical correlation | The company said an ongoing study correctly predicted patient response in 18 of 20 cases. No publication of that study was found. | Company-stated |
| Federal funding | No NIH, NSF or SBIR award to zPREDICTA found. The founder held an NIH R21 at Purdue in fiscal 2011. | Not found |
Read plainly: the science is real and it is in the literature, with named pharmaceutical co-authors and a large diagnostics partner willing to attach its name. What the record does not contain is the clinical validation the marketing rested on. The 18 of 20 figure was repeated for years and never published. The buyer paid $3.5 million for the technology and $200,000 for the customer relationships, a ratio that says it was buying a method rather than a book of business, then wrote both to zero inside a year.
What to watch
- Whether the PD-3D models still run anywhere. The subsidiary was absorbed at the end of 2022.
- The patent family. One application is abandoned, one grant is recorded, and no filing says who controls the estate now.
- Whether the FDA’s 2025 roadmap creates the market this platform was early for.
- Predictive Oncology itself, which raised about $343.5 million in October 2025 for a digital asset treasury.
In their words
“a clinical trial in a test tube.”
A customer, quoted in zPREDICTA’s YC application · Company
“Right now, 95% of new drug candidates that go into clinical trials to be tested in patients do not work in people.”
Julia Kirshner, co-founder and CEO, Propel(x) interview, 2017 · Interview
“It’s not enough to take cells and force them to clump into a three-dimensional structure. You have to provide them with the correct surroundings and the correct physiology for the resulting organoids to function properly.”
Julia Kirshner, Propel(x) interview, 2017 · Interview
“As this process is very close to what happens in a person, when the drugs are applied to this platform they behave in the same way as they do in a patient, resulting in a high correlation with clinical response.”
Julia Kirshner, Tecan Journal · Interview
“We believe the zPREDICTA 3D in vitro models offer a more predictable method for early evaluation of responses to immuno-oncology therapies.”
Steve Anderson, chief scientific officer, Covance, 2020 · Partner release
“zPREDICTA is the perfect fit for Predictive Oncology, providing yet another indispensable piece to the puzzle that will help humankind ultimately eliminate cancer,”
J. Melville Engle, CEO, Predictive Oncology, December 2021 · Company release
“It’s never dull, never the same.”
Julia Kirshner, Propel(x) interview, 2017 · Interview
Related companies
Sources
- Public recordPredictive Oncology Inc, Form 8-K on the zPredicta merger
- Public recordPredictive Oncology Inc, Forms 10-K for fiscal 2021, 2022 and 2023
- Public recordUS20190169562A1, Screening of immuno-modulatory therapies, zPredicta, Inc.
- Public recordMaytansinoid immunoconjugate IMGN901 is cytotoxic in a three-dimensional culture model of multiple myeloma
- Public recordFDA plan to phase out the animal testing requirement, and the Year 1 progress report
- IndependentZPredicta, LabCorp Form Cancer Cell Culture Model Partnership
- IndependentPredictive Oncology appoints Dr Julia Kirshner as its new chief scientific officer
- InterviewReconstructing Cancer Treatment: zPREDICTA Deal Profile
- InterviewTaking the guesswork out of drug development, and Don’t miss a beat with live cell imaging
- IndependentzPREDICTA company, funding and patent records
- CompanyPredictive Oncology Acquires zPREDICTA, Inc.
- CompanyzPREDICTA Y Combinator company page and application answers
Profile researched and written by Healthcare Discovery. Last updated September 29, 2026.
