☰

Cantargia AB

Running a virtual biotech on a shared data spine

Industry
Biotechnology
Headquarters
Lund, Sweden
Public information as of
January 2026

A4BEE prepared this analysis from publicly available sources. It reflects our own reading of Cantargia AB's published strategy and is not endorsed by, or produced in cooperation with, Cantargia AB. Company website

Strategic priorities

Cantargia AB is a clinical-stage antibody company targeting the Interleukin-1 Receptor Accessory Protein (IL1RAP) for oncology and inflammation. Lead asset nadunolimab (CAN04) is in development for pancreatic ductal adenocarcinoma (PDAC) and non-small cell lung cancer, with an FDA Fast Track designation for PDAC and a pivotal trial planned for mid-2026. The pipeline also includes CAN14 bispecific antibodies and IL1RAP-targeted antibody-drug conjugates (ADCs, antibody-drug conjugates — cytotoxic payloads attached to a targeting antibody).

The September 2025 closing of the CAN10 divestment to Otsuka Pharmaceutical — $33 million upfront and up to $580 million in milestones, plus a two-year right of first negotiation on next-generation IL1RAP antibodies — extends the cash runway into 2028 and provides the capital for new investment in digital infrastructure. The Q3 2025 interim report recorded net sales of SEK 308.7 million against an operating result of SEK 267.0 million, against 2024 net sales of zero.

Manufacturing is run as a virtual model. Industrial production of nadunolimab takes place at Thermo Fisher Scientific's Patheon Biologics site in Groningen, Netherlands, with earlier-stage manufacturing at BioInvent International AB in Lund and process development formerly at Celonic AG in Berlin. Clinical operations span MD Anderson Cancer Center, GEICAM, the Tisch Cancer Institute at Mount Sinai, and other academic and investigator-initiated collaborators, with a Chief Medical Officer who has published on the integration of radiology, pathology and genomic data into a "digital twin" of the patient's cancer.

The operational priority is therefore data integration: bringing the CDMO's bioreactor data, the academic clinical sites' case report forms, the Lund biometrics team's analysis environment, and the biomarker-stratification workflow that determines who gets enrolled into the pivotal trial onto one model that the leadership team can read end to end.

Challenges we see

  • Data Manufacturing

    Reading process signals from a partner's bioreactor

    Cantargia's lead asset is manufactured by Patheon Biologics in Groningen using cell culture followed by multi-stage purification. The CMC team sits in Lund, and production data moves between the two sites through reports rather than through continuous streams. Cantargia's annual report describes the production process as one that "takes place in cells" and that requires significant investment to make clinical development possible.

    Where manufacturing data arrives as a periodic report from the CDMO, the team sees the batch after it has been completed; instrumenting the boundary so process signals are read while the batch is still running narrows the window between a deviation and the moment it can be acted on.

  • Operations Regulatory

    Standardising biomarker scoring across PDAC patient biopsies

    Patient selection for the planned PDAC pivotal trial rests on Immunohistochemistry (IHC) scoring of IL1RAP expression, with patients split into "IL1RAP High" and "IL1RAP Low" groups based on the H-score (a semiquantitative measure of staining intensity and the proportion of stained cells). The CANFOUR trial classified 29 patients as High and 20 as Low, and high expression correlated with median overall survival of 14.2 months against 10.6 months.

    Where the H-score is read by eye against a reference image, two operators looking at the same slide can disagree by enough to move a patient across the threshold; capturing the score with a calibrated image-analysis model makes the threshold a property of the assay rather than a property of the reader.

  • Digital Integration

    Linking clinical sites to the Lund biometrics environment

    Clinical data is generated at MD Anderson, GEICAM, the Tisch Cancer Institute at Mount Sinai, and other academic and investigator-initiated sites under different data agreements and different case report form (CRF) schemas. The Cantargia biometrics team works on the integrated dataset once each study delivers.

    Where each trial delivers its own dataset, cross-trial comparison is a separate exercise for every question; aligning the schemas at the boundary lets the combined data be queried once and shortens the loop from a new signal to a protocol amendment.

  • Operations Operations

    Transferring process data when a programme changes hands

    The CAN10 programme was sold to Otsuka Pharmaceutical, with the asset purchase announced on 15 July 2025 and the transaction closing on 11 September 2025. Cantargia's development model deliberately drives programmes internally until a partner takes them, so tech transfer (the movement of process knowledge, documentation, and data from one organisation to another) is a recurring event.

    Where process knowledge lives in scattered records inside one organisation, the time it takes to share it with a new owner depends on how quickly those records can be packaged as a complete data set; structuring that data set from the start makes the next transfer a query rather than a project.

  • Compliance Regulatory

    Producing sustainability and regulatory documentation on a Nasdaq listing

    The Board has adopted a Sustainability Policy aligned with Sustainable Development Goals (SDG) 3, 5, 8, 9, and 13 and the European Union's Corporate Sustainability Reporting Directive (CSRD). Cantargia is listed on Nasdaq Stockholm and operates under FDA Fast Track designation for nadunolimab in PDAC, both of which carry reporting obligations.

    Where sustainability and regulatory data are aggregated by hand each cycle, the cycle itself sets the response time; capturing the underlying data once and assembling the reports from it turns the cycle into a routine rather than a peak-load event.

Opportunities, by urgency and business impact

Each bubble is one opportunity, numbered to match the list below. Further right means it bites sooner; higher means a bigger effect on the business. A bigger bubble means a bigger implementation effort.

Source: A4BEE analysis of public sources

  1. Reading process signals from a partner's bioreactor

    Cantargia's CMC team works from reports delivered by Patheon Biologics after each batch, so deviations in cell-culture conditions or downstream processing are visible only retrospectively and the link between a process parameter and a release decision has to be reconstructed.

    An agreed data contract between Lund and Groningen — what each bioreactor publishes, how it is timestamped, and how it reaches the CMC environment — gives the team a continuous view of every batch and lets quality signals be matched to process history from a single query.

    • Cantargia AB, About Nadunolimab (CAN04), 2025
    • Cantargia AB, Annual Report 2024

  2. Standardising biomarker scoring across PDAC patient biopsies

    Patient selection for the planned PDAC pivotal trial depends on an H-score derived by a human reader from an IHC slide, and the CANFOUR experience showed that reader-to-reader variability is large enough to move patients across the IL1RAP High / Low boundary.

    A calibrated image-analysis pipeline trained on the same reference slides produces a score with documented uncertainty, lets the same model be re-run when the assay is updated, and gives the regulatory dossier an auditable record of how each patient's score was generated.

    • Cantargia AB, Corporate Presentation, December 2025
    • Cantargia AB, AACR Conference Presentation on PDAC, 2025

  3. Linking clinical sites to the Lund biometrics environment

    Trial data from MD Anderson, GEICAM, the Tisch Cancer Institute, and other collaborators is delivered in different schemas and at different times, so cross-trial comparison is rebuilt for each question and the Lund biometrics team's response time is set by the slowest contributor.

    A shared data model covering patient, specimen, assay and outcome lets each site deliver into the same shape and gives the biometrics team a queryable, cross-trial view from which protocol-level decisions can be drawn without re-extracting the source data each time.

    • Cantargia AB, Business Development, 2025
    • Cantargia AB, First Patient Dosed in US Investigator-Initiated Colorectal Cancer Study, 2025

  4. Transferring process data when a programme changes hands

    The CAN10 programme moved from Cantargia to Otsuka in late 2025; future divestments and partnerships are a stated part of the development model, and each transfer requires assembling process records, analytical data and CMC documentation from systems that were not designed to be packaged for an outside party.

    A documentation backbone that stores process and analytical data in a structured form from the start makes the next transfer a data export against an agreed schema rather than a project to reconstruct the history under deadline.

    • Cantargia AB, Press Release on CAN10 Acquisition by Otsuka, July 2025
    • Cantargia AB, Press Release on Closing of CAN10 Acquisition, September 2025

  5. Producing sustainability and regulatory documentation on a Nasdaq listing

    CSRD-aligned sustainability reporting, FDA Fast Track data integrity commitments, and Nasdaq Stockholm listing requirements each demand their own document assemblies, and the same operational data has to be re-told in each format on each cycle.

    A sustainability and regulatory data layer that captures the source numbers once and assembles the CSRD report, the Fast Track data package, and the listing disclosures from the same store replaces three parallel manual assemblies with one source of truth and a small set of formatted outputs.

    • Cantargia AB, Integrating ESG for a Sustainable Future, 2025
    • Cantargia AB, Interim Report Q3 2025

What we'd propose

  • Digital CDMO

    Connected data contract with the Patheon manufacturing site

    An agreed OPC UA (Open Platform Communications Unified Architecture) or MQTT (a lightweight publish-subscribe messaging protocol commonly used for sensor data) interface between Patheon Biologics in Groningen and the Cantargia CMC team in Lund that publishes each bioreactor's process parameters in real time and routes them into a per-batch record at Cantargia.

    • Per-batch data contract

      What each run publishes

      Define the tags, sampling rate and timestamp format for cell-culture and purification parameters that leave each Patheon run, so the Lund team reads the same shape every batch regardless of the equipment revision in use.

    • Edge gateway at the CDMO boundary

      A documented translation layer

      Place a small gateway at the Patheon interface that translates vendor-specific protocols to the agreed data contract and signs each value so the receiving system can verify provenance without needing direct access to the controller.

    • Continuous view in the CMC environment

      Batches seen while they run

      Stream the contracted tags into the CMC team's batch record so deviations are visible against the run that is in progress and release decisions are read from the same history the engineers see.

    • Process deviations are visible against the batch that is in progress, not the one that already shipped.
    • The CMC team works from the same history as the CDMO engineers, so the release decision sits next to the evidence behind it.
    • Future partnerships inherit a documented data interface rather than starting again from raw exports.

    We've done this before

    See how we do Digital CDMO
  • Digital Lab

    Computer-vision scoring for IL1RAP expression on PDAC biopsies

    A calibrated image-analysis pipeline that takes an IL1RAP IHC slide as input and returns an H-score with documented uncertainty, replacing the manual read while leaving the pathologists in the approval loop.

    • Scoring model calibrated to the reference set

      Trained on the CANFOUR slides

      Train the model on the same reference set the CANFOUR analysis used, so the model's output is traceable to the dataset that established the IL1RAP High / Low threshold and the documented survival benefit.

    • Auditable score with uncertainty

      Every score carries its own evidence

      Return the H-score together with a confidence interval and a heatmap of the regions that drove it, so a reviewer can read the model's reasoning and the dossier can show how each patient's score was produced.

    • Regulated re-run and version control

      The assay stays reproducible

      Version the model and the reference set together, so a re-run on a stored slide produces the same score and an updated assay can be re-evaluated against historical cases without leaving a gap in the audit trail.

    • Reader-to-reader variability stops moving patients across the IL1RAP High / Low boundary.
    • The pivotal trial is enrolled against a single, reproducible assay whose history is auditable end to end.
    • The same scoring pipeline is available at every clinical site without needing the same pathologist in the room.

    We've done this before

    See how we do Digital Lab
  • Enterprise AI

    Shared clinical data model across MD Anderson, GEICAM and Tisch Cancer Institute

    A shared data model for patient, specimen, assay and outcome that each collaborator's case report form feeds into, so the Cantargia biometrics team reads one integrated dataset rather than three separate ones.

    • Common ontology for trial data

      One agreed set of terms

      Define patient, specimen, assay, indication and outcome as explicit entities with agreed relationships so a query written once returns comparable answers across MD Anderson, GEICAM and the Tisch Cancer Institute instead of three dialects of the same table.

    • Per-site ingestion with provenance

      Each site delivers into the same shape

      Build an ingestion path for each collaborator's case report form, schema-validated at the boundary so a record that does not fit the model is rejected with a clear message rather than silently joining the dataset in a degraded form.

    • Cross-trial analytics on the integrated view

      One query, one answer

      Expose the integrated dataset through dashboards and a retrieval layer so the biometrics team can answer cross-trial questions directly and the clinical operations team can see enrollment and biomarker signal in the same place.

    • Cross-trial questions are answered against one dataset rather than rebuilt for each study.
    • Protocol amendments are read from a continuously updated view of the data rather than from the most recent site delivery.
    • New collaborators attach to the shared model rather than triggering another round of data harmonisation.

    We've done this before

    See how we do Enterprise AI
  • Digital Lab

    Structured documentation backbone for tech transfer

    A documentation backbone that stores Cantargia's process knowledge, analytical data and CMC records in a structured form from the day they are generated, so the next partnership or divestment reads from a complete data set rather than a reconstructed one.

    • Structured process records

      Documents that are data

      Capture process recipes, analytical results and batch outcomes as records in a structured store with version history and explicit relationships, so the dataset that supports a future transfer is generated by the work itself rather than assembled for it.

    • Transfer package assembly

      A complete export on demand

      Generate a transfer package against the agreed data contract when a programme moves, with the same provenance and audit trail the original records carried, so the receiving party receives what the original site would have shown an inspector.

    • Change tracking across the documentation set

      Every change is findable

      Index every controlled document against the underlying records it references, so a query about a specific process change returns every document the change affected and the history of those documents.

    • Each transfer is an export against an agreed schema rather than a project to reconstruct history.
    • The receiving party inherits the same evidence base the original site had, so due-diligence questions are answered from the record.
    • Records accumulated for one transfer become reusable the next time, lowering the marginal cost of each partnership.

    We've done this before

    See how we do Digital Lab
  • Agents

    AI agents for sustainability and regulatory documentation

    Narrow AI agents that draft the recurring parts of sustainability reporting, FDA Fast Track data packages and Nasdaq listing disclosures from the same source data store, with a named reviewer approving every output before it leaves the organisation.

    • Drafting from source records

      First drafts from the data

      Generate the first draft of a CSRD-aligned sustainability section, a Fast Track data package or a Nasdaq listing disclosure from the same structured store the operational systems feed, so the author edits and judges rather than assembles.

    • Template and completeness check

      Gaps caught before review

      Check a draft against its template and the checklist Cantargia's own quality system uses, returning missing or inconsistent sections before the document enters the human review queue.

    • Standards-change impact search

      Which documents a change touches

      When a sustainability standard, a Fast Track reporting requirement or a Nasdaq disclosure rule changes, retrieve every controlled document the change affects and rank them by how directly the change applies, so the update scope is known on day one.

    • Three parallel manual document assemblies are produced from one source of truth.
    • Review queues move faster because documents arrive complete and against the right template.
    • A change to an external standard is mapped to the document set immediately, before the disclosure deadline sets the response time.

    We've done this before

    See how we do Agents

Digital maturity: today and target

Scored out of 100 across six dimensions. The target is what Cantargia AB's own published ambition implies — not a perfect score.

Source: A4BEE analysis of public sources
Manufacturing data integration 35 → 80
Manufacturing sits with Patheon Biologics in Groningen and the CMC team in Lund reads each batch from a report. An agreed data contract between the two sites would close that gap and give the team the same view of the batch that the engineers on site have.
Clinical data analytics 40 → 80
MD Anderson, GEICAM and the Tisch Cancer Institute deliver under different schemas and on different schedules, so the Lund biometrics team works on a per-study basis rather than on an integrated dataset.
Biomarker automation 25 → 85
The CANFOUR IL1RAP H-score was read by eye and reader-to-reader variability is documented in the publication; a calibrated image-analysis pipeline would replace that read with a reproducible assay.
R&D data platform 45 → 80
The CANxx antibody library carries binding and cytokine-blocking profiles that could feed a candidate-prediction workflow if the records were available in a structured form rather than across documents and spreadsheets.
Regulatory data integrity 50 → 85
FDA Fast Track designation and Nasdaq Stockholm listing both demand evidence generated continuously rather than assembled at filing time; the current documentation work is largely manual.
IT/OT convergence 30 → 80
The CDMO boundary in Groningen and the academic clinical sites do not connect to the Lund environment in real time, so any data exchange is a discrete project rather than a continuous flow.

Check this yourself

Our Service Portal has free self-assessments and market comparisons. These are the ones that line up with what we've read above — no sales call required.

Think we've read this right?

Talk to us

Related reading

This is an independent analysis prepared by A4BEE from publicly available information as of January 2026. It reflects A4BEE's own interpretation and opinion, is not affiliated with, endorsed by, or verified with Cantargia AB, and may be incomplete or inaccurate. All company names and trademarks are the property of their respective owners. To request a correction or removal, contact [email protected].