Curevac’s credibility unravels along with its Covid-19 vaccine

Long-awaited results from the Herald study are a bust, and the data will likely deteriorate further.

There is no getting away from the harsh reality of Curevac’s long-awaited phase 2/3 readout yesterday: CVnCoV, its latecomer Covid-19 vaccine, is not competitive. An interesting question, however, is why these second interim analysis data were disclosed at all, and goes to the heart of Curevac’s fast-shrinking credibility.

Indeed, as recently as one week ago the company was insisting to the sellside that the second interim analysis had not taken place. This is by no means the only questionable aspect of CVnCoV’s development; perhaps given the confusion, and how clearly the vaccine was underperforming, Curevac had no option but to reveal all.

The data announced yesterday concern CVnCoV’s Herald study, carried out mostly in South America and at some European sites. Its readout had been keenly awaited, given CVnCoV’s status as potentially the third mRNA-based Covid-19 vaccine, and lack of news had spurred speculation that stopping criteria were being missed for lack of efficacy.

Interim analyses

Finally Curevac revealed on May 28 – after market close on the Friday before a US holiday – that a first interim analysis had come and gone without meeting efficacy criteria.

Yesterday brought news of the second interim analysis, which yielded 47% vaccine efficacy among a total 134 adjudicated Covid-19 cases. This is clearly disappointing, underperforming the likes of Moderna and Pfizer/Biontech’s 95% figures, and the EMA’s loose approval threshold of 50%; Curevac crashed 48% in early trade today.

The data raise several questions. The Herald study’s updated statistical analysis plan specified efficacy of around 44% as the threshold below which futility would be declared at second interim; actual efficacy has beaten this, albeit marginally, but clearly the 60% threshold for success has not been breached either.

So why even announce these data? Standard practice would be to say stopping criteria had not been reached, and proceed to final analysis.

There are two possibilities: either Curevac has effectively given up on Herald, or it felt that it had to reveal the data given its earlier pronouncements. These had included a robust denial of a report in Augsburger Allgemeine, a newspaper that had claimed that the second interim analysis had, like the first, come and gone.

But on June 9 Evercore ISI analysts had reported Curevac as “categorically denying” the newspaper’s claim, writing: “second interim ... has not been done”. Yesterday’s announcement shows that the Augsburger Allgemeine was on the right track.

This piles more questions on a project that had been mired in controversy right from the start, when a political storm over an alleged exclusive US deal ended with the abrupt departure of Curevac’s chief executive (Crisis, what crisis? Curevac denies everything, March 17, 2020).

And things get worse still. Curevac has said that Herald’s second interim analysis ended up looking at 134 Covid-19 cases – which, remarkably, is under a third of the total cases there have been so far in the study; 474 further possible cases have been noted, but these are still subject to adjudication.

Stopping criteria for Curevac's Herald study
  Interim analysis 1 Interim analysis 2 Interim analysis 3
Cases needed to trigger 56 111 160
Cases adjudicated 59 134 ~215*
Efficacy threshold for success ≥80.9% ≥59.5% ≥50.5%
Efficacy threshold for futility ≤19.4% ≤43.7% NA
Actual efficacy Not disclosed 47% NA
Result Study continued Study continued NA
Possible cases as yet unadjudicated Not disclosed 474 NA
Source: trial protocol & Curevac press releases. Note: *Curevac guidance.

Curevac’s defence centres on the growing prevalence of Covid-19 variants of concern, which clearly affected Herald to a greater extent than competitors’ early Covid-19 trials. On a call today management said the Covid-19 landscape had “changed dramatically”, meaning that it was now “virtually fighting a different virus”.

This reflects the way any infectious disease adapts, and highlights the difficulty of being a relative latecomer like Curevac. But many other groups continue reporting reasonable efficacy against some emerging variants, especially as regards hospitalisations, a subject on which Curevac has so far been silent.

Then there is the problem of Peru. Though Curevac carefully avoided splitting out details, Evercore estimates that CVnCoV’s efficacy against the C.37 Peru variant could be only 20%. This is crucial, as this variant accounts for most of the 474 currently unadjudicated cases; thus Curevac’s dataset is likely to get worse as adjudication of Peru cases swings the dataset further towards failure.

For now Curevac is saying Herald will proceed to final analysis. It is not clear whether a Europe filing will take place, but the company stressed its move to second-generation Covid-19 vaccines, which it claims cause higher protein expression and have improved kinetics.

This work is the result of a deal with Glaxosmithkline. The UK big pharma group has notably missed the boat in Covid-19, and Curevac’s data are a further example of it having hitched its wagon to the wrong horse.

Selected large trials of Covid-19 vaccines
Company Vaccine Study Headline efficacy Variant detail
Biontech/ Pfizer Comirnaty C4591001 (mostly US) 95% ~88% efficacy vs India variant
Moderna mRNA-1273 P301 (US) 95% ~88% efficacy vs India variant
Johnson & Johnson JNJ-78436735 Ensemble (mostly US; single dose) 67% 66% efficacy vs Brazil variant, 52% vs S Africa variant
Astrazeneca AZD1222 NCT04516746 (mostly US) 76% (62 vs 128 for control)* ^ 60% efficacy vs India variant; S Africa variant caused efficacy to fall to 10% for mild disease
Novavax NVX-CoV2373 Prevent-19 (mostly US) 90% 50% effective vs S Africa variant; 93% against variants of concern/interest
Curevac CVnCoV Herald (75% S America, 25% Europe) 47% (46 vs 88 for control)* ^^ Efficacy not split out, but cases mostly Kent & C.37 Peru variants; 370 Latin America cases yet to be adjudicated, incl ~160 in Peru (~100 C.37 variant) and ~100 in Argentina (~25 C.37 variant)
Source: company announcements, FDA documents & scientific publications. Note: *absolute numbers estimated; ^2:1 randomisation, and a further 14 cases yet to be adjudicated at 25 Mar 2021; ^^a further 474 cases yet to be adjudicated at 17 Jun 2021; Curevac competitors have also cited continuing protection vs severe Covid-19 and hospitalisations due to novel variants, though there is no data on the C.37 Peru variant.

Share This Article