The day after data appeared from the vaccine maker Sinovac showed complete protection of rhesus monkeys by their vaccine candidate (whole inactivated SARS-CoV-2 virus particles), scientists from the Jenner Institute in Oxford issued a press release announcing that their vaccine (an adenovirus vector based vaccine that carried the SARS-CoV-2 spike protein) worked to protect rhesus monkeys and that they were moving forward with large scale human safety trials. At the time, the substantiating data was not available. Now it is, in the form of a May 13 BioRxiv preprint. Does the data support the claim?

Not really. All of the vaccinated monkeys treated with the Oxford vaccine became infected when challenged, as judged by recovery of virus genomic RNA from nasal secretions. There was no difference in the amount of viral RNA detected from this site in the vaccinated monkeys as compared to the unvaccinated animals. Which is to say, all vaccinated animals were infected. This observation is in marked contrast to the results reported from Sinovac trial. At the highest dose studied, no virus was recovered from vaccinated monkeys from the throat, lung, or rectum of the vaccinated animals.

There is a second troubling result of the Oxford paper. The titer of neutralizing antibody, as judged by inhibition of virus replication by successive serum dilutions as reported is extremely low. Typically, neutralizing antibodies in effective vaccines can be diluted by more than a thousand fold and retain activity. In these experiments the serum could be diluted only by 4 to 40 fold before neutralizing activity was lost. Again, by contrast the titer of neutralizing antibodies in the serum of those vaccinated with whole inactivated SARS-CoV-2 was high.

What then is the argument for pressing forward with the adenovirus vector SARS-CoV-2 spike protein vaccine?

The authors present evidence to the effect that, although the vaccine did not protect the animals from infection, it did moderate the disease.

The vaccinated and control animals were followed for clinical signs of infection for seven days post infection. One clinical sign of infection in rhesus monkeys is breathing rate. Monkeys ill from SARS-CoV-2 infection breathe more rapidly than normal. By this measure 3 of the 6 vaccinated monkeys were clinically ill, the remaining three were not clinically distinguishable from the unvaccinated animals.

A second test is measurement of the amount of virus in the lungs (bronchial lavage). Viral RNA was detected in the bronchial lavage of 2 of the 6 vaccinated animals and in all three unvaccinated animals, again suggesting only partial protection.

At day seven post challenge the animals were euthanized and examined for lung damage. Two of the three unvaccinated animals “developed some degree of interstitial pneumonia” as judged by pathological examination of the lungs. No such damage was observed in any of the vaccinated animals.

It is encouraging that no evidence of vaccine induced disease enhancement was observed in either the Sinovac vaccine nor the Oxford trials. However, experience with other vaccines tells us that is not a firm guarantee that such will be the case for humans.

What to make of this data? It is crystal clear that the vaccine did not provide sterilizing immunity to the virus challenge, the gold standard for any vaccine. It may provide partial protection.

The question then becomes: Will partial protection be enough to control the COVID-19 pandemic? That is an open question. For an answer we can look to other diseases for which only partially effective vaccines exist—HIV, tuberculosis, and malaria. The answers are not encouraging, except perhaps for the protection of childhood malaria.

What are the potential implications for other vaccine trials? We know adenovirus is a good vector for eliciting protective responses for protein coding genes of other viruses, in fact better than many others. This then raises the question of whether vaccine strategies based on the delivery of viral antigens by nucleic acids, be it DNA or RNA, will suffice. From the published data the nucleic acid of select viral genes and proteins delivery technology seems inferior to a whole killed virus vaccine approach.

What then are the choices for the Oxford group? Steam ahead with a vaccine known to be partially effective at best, one which we already know elicits poor neutralizing responses? Work to improve the immunogenicity of the current vaccine? Attempt trials with combinations of two or more vaccines, a prime boost strategy?

These questions are all the more fraught by what we already know about the complexities of antibody reposes to natural infections. For example, we know in the case of SARS and other coronavirus infections that even high titers of neutralizing antibodies fade quickly over time. How long can we expect weakly neutralizing antibodies to protect?

We know what the public response is of the Oxford group and their collaborators. Proceed with full speed to human safety then efficacy trials. Time will tell if this is the best approach. I wouldn’t bet on it.

This article originally appeared on Forbes