NorthWest Biotherapeutics Inc (NWBO)

[biosectinvestor]

Correction: Why NWBO’s “other-trial” control arm still sits squarely inside the MHRA draft guidance


The actual guidance from the MHRA says:

While the guideline is specifically aimed at sponsors planning to use RWD ECAs, many of the general principles would be relevant for external controls drawn from other sources, such as previously completed clinical trials. MHRA is also interested in engaging with sponsors who have proposals for using such data sources.

1. The guidance expressly covers trial-derived datasets
Paragraph 8, quoted above, clearly indicates their willingness to accept data from clinical trials, and in this case, concurrent clinical trials using the same general SOC.

2. Concurrent controls are “ideal”; historical controls are an accepted fallback
• Paragraph 29: an external arm “can be concurrent or historical; concurrent data collection is considered the ideal.”
• Paragraph 35 repeats that concurrent controls are generally preferable because they avoid time-trend bias.
• Paragraph 55 allows historical controls when no fit-for-purpose concurrent registry exists.

3. NWBO’s comparators meet that “ideal” definition
The SAP required every control study to follow the Stupp standard of care and to enrol patients in the same therapeutic era as the DCVax-L cohort. Five newly diagnosed and ten recurrent GBM RCTs met those rules; JAMA Oncology calls them “contemporaneous, matched external control patients.”

4. Data quality outweighs provenance
Paragraph 16 prioritises “quality and fitness-for-purpose” over where data come from. Completed GBM RCTs provide adjudicated endpoints, uniform imaging schedules, and granular baseline variables—qualities most routine-care databases lack.

5. NWBO applied every bias-mitigation tool the draft asks for
• Pre-specified selection and analysis in a locked SAP.
• Reconstructed individual-patient data (Guyot algorithm) wherever raw IPD were absent.
• MAIC weighting to align baseline age, sex, MGMT, KPS, residual disease, etc.
• Six leave-study-out sensitivity analyses—hazard ratio stayed 0.77-0.82.
• “Validator-switch” tests showed the method does not create false positives.

6. The dataset is clearly “external” to the DCVax-L trial
Paragraph 14 defines RWD as data “collected outside of a clinical study” under review. The comparator trials concluded years before DCVax-L unblinding, so their control arms are indeed external.

7. No appropriate registry exists—so the guidance points to exactly this solution
Temozolomide-era GBM registries with fully imaged concurrent controls don’t exist. Under paragraph 55 a rigorously matched historical (or other-trial) solution is the next step—NWBO surpassed that bar by choosing concurrent RCT controls instead of older historical cohorts.

Conclusion:
The draft MHRA hierarchy is clear: concurrent registry controls first, concurrent RCT controls next, historical controls last. NWBO stands firmly in tier 2—contemporaneous RCT controls treated under the same SOC as DCVax-L patients—and then layers reconstructed IPD, MAIC weighting, and multiple robustness checks. For a rare, lethal disease where randomising more patients to placebo is unethical, this package delivers exactly the bias-mitigation blueprint the draft guidance prescribes.