20 Dec Reflections on phase 3 trials and early roll-out of COVID-19 vaccines

Sun, 12/20/2020 - 20:41

Episode 95: Reflections on phase 3 trials and early roll-out of COVID-19 vaccines

Dear colleagues,

After reading the papers/reports on the phase 3 trials of the forerunners and some additional comments, I’d like to share my provisional conclusions and invite you to share your reflections and reactions as well

  1. My first impression is that both mRNA vaccines (BNT162b2 Pfizer-BioNTech and mRNA1273 Moderna) are equivalent in many respects (> 90 % efficacy and, at first view, acceptable side effects).  (see Ep 95 1 and 2 FDA reports and Ep95 3a and b NEJM paper on Pfizer vaccine)

A number of our previous concerns have been met:

  • Primary efficacy against typical COVID (at least 1-2 weeks after second dose) is similar across age groups (also > 65, and even > 75) , genders, racial and ethnic groups, and participants with medical comorbidities (including, pulmonary, cardiovascular, diabetes,  obesity).
  • Secondary efficacy suggested benefit in preventing severe COVID-19, and in preventing COVID-19 following the first dose.
  • Besides very notable local reactions, systemic grade 3/4 severe reactions were not uncommon in the vaccine group, especially after the second dose and included headache, fatigue, chills, myalgia, arthralgia and fever, but these reactions usually subsides without much treatment after a few days.   
  • There was no evidence of anything like “antibody-mediated enhancement” in vaccinated subjects that got infected anyway and subjects with pre-existing SARS-CoV-2 antibodies experienced rather less side effects.  
  • Older subjects tended to suffer rather less side effects.


  • Whether or not COVID-19 can be prevented in individuals with prior SARS-CoV-2 infection, remains less clear, because of rather numbers of people with positive serology included (still many hundreds), but across both trials only 3 reïnfections (or relapses?) were documented.  
  • Similarly, numbers of people with HIV infection or chronic liver disease were too low to make conclusions.
  • In both trials, it is really the disease COVID-19, which is prevented: both mild and severe COVID-19, but systematic screening for (asymptomatic) infection and transmission has not been done.
  • Although COVID-19 was prevented in both “healthy” and “comorbid” older subjects, the most vulnerable “frail” elderly were not represented.  Whether they are protected will become evident during roll-out, because they are now among the first to be vaccinated.
  • We have no insight in the correlation between induced immunity (neut antibodies) and efficacy.

Unresolved questions that will be part of pharmacovigilance:

  • Protection against asymptomatic infection, infection in “frail” elderly, pediatric patients, pregnant women, in various high risk populations with HIV, complicated co-morbidities etc, in previously SARS-CoV-2 infected individuals…
  • Duration of protection also against severe disease and mortality
  • Escape of virus from vaccine….

Some further practical comparisons can be found in Ep 95 4.

  1. The Lancet paper on the AstraZeneca Oxford ChAdOx1 nCoV-19 vaccine (Ep 95 5) is much less convincing to me.  Once you start reading the “methods” section, you realize that the phase 3 trial has been a very complex and painful process.  Some problems that occurred:
  • The vaccine has been produced by two different manufacturers: first Advent (Italy) and later by COBRA Biologicals in UK.  It remains unclear to me if this (unforeseen?) switch has contributed to logistic and standardization problems (below).
  • The time between priming and boosting is not constant: in particular cohorts longer intervals were used: the target was 4 weeks, but it could be as long as 12 weeks.
  • In some cohorts a reduced dose was used as prime, which, at first view could be fine, but it becomes really complicated when you read the explanation on p. 3, where you learn that the definition of the dose has “evolved” over time
  • Honestly, I’m not sure about standardization of doses over time and across sites.

To me, it comes as no surprise that the results, shown in Table 2, are not all very convincing.  We learn that for the “primary outcome” (typical COVID symptoms, confirmed with PCR) the overall efficacy in the subjects, who received two standard doses (SD) is 62 % (CI 41-75) and for any PCR positivity 57 % (41-66). The “90% efficacy” of the press communication was found in a rather small subset (2,700 subjects) who received a low dose (LD) first and a SD as boost, but it has also a wide CI (67-97 %).  

Table 3 suggests that it is better to have a larger interval between the doses: efficacy increases from 53 to 65 % if interval longer than 6 weeks. 

In addition there is a small sub-study in UK, where self-swabbing was done and which could inform us about efficacy against asymptomatic infection.  The text on p.8 reads as follows: “Asymptomatic infections or those with unreported symptoms were detected in 69 participants. Vaccine efficacy in the 24 LD/SD recipients was 58,9% (95% CI 1,0 to 82, 9), whereas it was 3,8% (−72,4 to 46,3) in the 45 participants receiving SD/SD (table 2)

There is no data on older subjects here, some people with comorbidities are included, but there is no specific analysis on those….


  1. Additional information on Pfizer and other vaccines:


  • The NEJM Editorial, accompanying the Phase 3 Pfizer trial (Ep 95 6) summarizes some of the concerns, we have discussed above, including possible bias in recording symptomatic and asymptomatic infections, questions on logistical problems (storage at -70, need for 2 doses), long-term efficacy and occurrence of unexpected side effects.


  • And indeed, as we know, the Pfizer vaccine has induced at least 3 anaphylactic shocks in HCW since it is being rolled out.  This could happen with any vaccine, especially in people with a history of allergic reactions (which was the case for the first two, but not the third “Alaska” case). Cabanillas et al. propose that poly-ethylene glycol could be the culprit.  If so, it will also occur with the Moderna vaccine  (Ep 95 7 A, B, C).  This hypothesis has also been proposed by the German Allergology Societies (Ep 95 D).  They formulate an advice that I have translated (Ep 95 7 E)


  • After the relatively good news on the Chinese inactivated vaccines, there is now a European company (Valneva) that also aims to develop such a “simple traditional” vaccine:  the main difference with similar Chinese (and Indian?) inactivated vaccines being that besides Al-OH, also CpC is added as a second adjuvant that could induce a stronger  Th1 cellular response than the existing ones. (Ep 95 8).


  • There is also NVX-CoV2373 protein vaccine (Novavax USA): it is a full prefusion stabilized trimeric S in a nanoparticulate  formulation with Matrix-M saponin-based adjuvant.  Results in cynomolgous macaques show that a 5 µg dose twice with adjuvant induces solid immune responses and actual protection (Ep 95 9A) and the phase 1-2 in humans confirms induction of very solid neutralizing antibodies and Th1 responses, with either 5 or 25 µg but with the adjuvant.  Side effects seem rather mild to moderate and are clearly depended on the adjuvant (Ep 95 9B).  In the discussion, the authors claim that this adjuvant has already been applied in more than 14,000 participants in various nanoparticle vaccine trials, including children, pregnant women, and older adults”. They also promise that older people and those with comorbidities will be included in phase 3, which is apparently ongoing in the UK (Ep 95 9C).   


Best wishes,