Updated Apr 25, 2021: Additional data from Frantzen et al NDT 2021; Rincon-Arevalo et al, MedRxiv 2021; 8 week data Goupil et al MedRxiv 2021.
Updated Apr 29, 2021: Additional data from Jahn et al Vaccines 2021; Chavarot et al Transpl 2021 and Yi et al Transpl 2021
Updated May 14, 2021: Additional data from Boyarski et al JAMA 2021 [NB - there are two Boyarski JAMA publications] and Anand et al MedRxiv 2021. A few tweaks to the text. I’ve written a similar piece for Kidney Care UK, answering frequent questions they receive from patients - click here if you want a more patient-orientated piece.
Over the last two months many studies have been published, or pre-printed, regarding COVID-19 responses in haemodialysis and transplant patients. The headline results can look very different between studies and can look pretty bleak. The idea here is to provide some context from which to assess these studies. All of these studies look very early – the latest is 7 weeks after starting a vaccination course – and so interpreting these results needs a little thought about what antibodies do very early after vaccination. The majority of the studies report spike (or its RBD subunit) IgG antibodies, so we will focus on interpreting those results.
What is the timeframe of anti-S antibodies in the healthy population?
Walsh et al. NEJM Dec 2020 reported the anti-S1 antibody titres for the now licenced 30micrograms of BNT162b2 at days 14, 21, 28 and 35. An abridged version of their figure is here (showing just the 30mcg BNT162b2 data):
Figure from Walsh et al, NEJM 2021
The antibody response approximates to an S-shaped curve – there is a little by day 21 and then an explosion of antibody in serum (note log scale). It is easier to see this pattern with weekly early timepoints using data from Xia et al. Lancet Infectious Diseases (a developmental subunit vaccine, rather than BNT162b2).
Figure from Xia et al, Lancet Inf Dis, 2021
This makes immunological sense as the sequestration of antigen in the draining lymph node and the careful assembly of the co-ordinated and tightly controlled germinal centre reaction takes a few days, before it begins to produce antibodies (and B cell memory) of iteratively higher affinities. All immunisations look like this early on. It is worth noting that older individuals have a right shift in their S curve, with a response that’s a little slower to get going.
In summary, you’d expect to see antibodies at around 2-3 weeks in younger healthy individuals, and around 3-4 weeks in older, healthy individuals.
What is the timeframe of antibody generation in dialysis patients?
The Attias et al. study in Kidney Int measured antibody titres every week, which allows plotting like the phase II trials above.
Figure from Attias et al, Kidney Int 2021
For SARS-CoV-2 naïve patients (in purple), the now familiar S-shape response is seen again. Attias et al. have a break, and scale change, that squashes the S down. These patients only really start to get an antibody response going at weeks 5-6, a week or so later than the friskiest of 65-85 yo individuals in the BNT162b2 phase II (the Attias average age was 70 in their dialysis cohort). This looks like dialysis patients are right shifted again compared to (roughly age matched) healthy controls. The HD studies report as a slice in time through these S curves eg at week X the HD cohort had less antibody than the control cohort (some of the studies have age-matched controls, some do not – this is a risk when doing research at full-pelt in a pandemic).
Nineteen percent of the Attias HD cohort had prior COVID-19 infections – these are plotted in blue, and their curves are very left shifted. These individuals are mounting a secondary response, and a larger, quicker response is entirely predictable, as part of the T and B cell components needed for the germinal centre have already been trained.
It is worth noting the variation you can see in the Attias plot too. There is a lot. That’s typical of human vaccination response measures (or any human immune parameter you care to measure – look at the ‘normal’ range for lymphocyte count next time you look at differential). The variation is biggest around the slope of the S (day 35), so if a study makes a comparison at that sort of time, then it will be measuring something at a point where there is lots of natural variation. Most of the transplant studies (see later) are around this time.
We can see that the timing of antibody sampling is crucial – very early will be the flat part of the S curve, later will be the rise and even later in time will be the plateau. If you plot out the 11 studies so far, you can see this sort of relationship seems to hold true for patients on HD (several caveats coming next):
Caveat 1: The gold-standard plot would use the primary data from each study. The Attias study data would be very useful as there are no datapoints above for weeks 1-3. I’ve added the origin as a datapoint to fit an S curve.
Caveat 2: Antibody measures and how you can compare them between studies. Nearly all studies use a different kit to measure antibody – these are not directly comparable – unlike different FBC / CBC kits where the Hb would be nearly identical, excepting units, across the globe. A way to generalise is to use the % responders (i.e. those with any detectable antibody); a better way would be to run a set of same samples across the whole dynamic range on each assay. If you explore which assay(s) were used in each study you can see a hint of some trends, for example both Simon et al. and Torregiani et al. shared the same assay, and tend to sit a little lower than the fit line, but it’s very difficult to make firm conclusions given the 9 different assays used across first 14 studies (May, 14: not yet updated with latest studies/assays, but the heterogeneity remains):
Sources: Goupil et al, MedRXiv 2021; Simon et al, MedRxiv 2021; Schrezenmeier et al MedRxiv 2021; Grupper et al, cJASN 2021; Agur et al, NDT 2021; Sattler et al MedRxiv 2021; Lacson et al MedRxiv 2021; Torreggiani et al, Kidney Int 2021; Attias et al Kidney Int 2021; Billany et al Kidney Int 2021; Berar Yanay et al Kidney Int 2021; Jahn et al Vaccines 2021
Caveat 3: Uncertainty along both axes and around fit line. I’ve summarised to the nearest half-week, and I’ve not shown error bars for the y axis of any study, or the fitted line.
Caveat 4: Not all studies had dose 2 at day 21 (hence small dash line at 3 weeks), a handful did day 28 dosing. There are three (Toreggiani, Goupil and Billany) which report after first dose only. Note 4 and 8 week data from Goupil et al show little additional immune response with time.
Caveat 5: prior SARS-CoV-2 exposure. As much as possible from the manuscripts I’ve tried to extract data for the primary, SARS-CoV-2 naïve patients. This is not always possible. Clearly if a few exposed/immune patients have snuck through it will change things.
Caveat 6: What sort of antibody is measured? Some of the assays measure IgM, IgG or IgA individually or as a single number reflecting their combination. The S curve for IgM will be earlier than IgG and IgA. As much as I can (caveat 1 + 2), I have tried to plot the IgG data, as that’s the predominantly tested single isotype.
Caveat 7: Weighting. It would be sensible to weight each study by its size – the numbers of HD patients in each study is plotted above. This S curve does not add weight to each study.
Are dialysis patients protected?
The majority of dialysis patients make at least some antibodies by week ~7. It is reasonable to expect these antibodies to protect from, or attenuate disease, but it is not formally proven.
A lab assay that narrows that knowledge gap is micro-neutralisation titres against live virus (and its variants) which tests for direct neutralisation activity of serum. Only a few labs do it: you need high levels of biosafety, technical expertise, and the ability to do it at scale (100s of sera / day). We have this planned, as research, for UK dialysis patients.
What about transplant recipients?
There are fewer transplant studies. We have plotted them against time here – all of the caveats above still apply. The plot now uses colours reflect whether the study is HD or transplant and whether it was a single or two dose studies. The size of each point reflects the number of HD or transplant patients with data at the relevant timepoints. The first thing to highlight is that the vaccines (predominantly BNT162b2 in these studies), do something – they may not be perfect, or work as quickly as ideal, but they look quite a lot better than doing nothing. Second, there are no data yet available after week 9. It is very difficult to predict what the antibody titres will do after week 9. Using the same cohorts there is an increase over time: Sattler et al. report two timepoints in 1 study and there are two Benotmane et al. publications – one for each timepoint (in Benotmane #2 there are slightly fewer patients who’d reached week 7 at time of publication). Chavarot et al also report a marginal increase over time. Whilst the early antibody responses (within 8 weeks) are modest, these longitudinal re-sampling studies offer perhaps some optimism. Have the current studies just caught the start of the up-tick of the S curve? Or has it already reached a plateau? There’s not enough data here to draw even the heavily caveated HD-style S. Data from weeks 8, 9 and 10 and beyond will help. Clinician vigilance for reporting post-vaccination infections has a role too.
Sources:
HD Studies: Goupil et al, MedRXiv 2021; Simon et al, MedRxiv 2021; Schrezenmeier et al MedRxiv 2021; Grupper et al, cJASN 2021; Agur et al, NDT 2021; Sattler et al MedRxiv 2021; Lacson et al MedRxiv 2021; Torreggiani et al, Kidney Int 2021; Attias et al Kidney Int 2021; Billany et al Kidney Int 2021; Berar Yanay et al Kidney Int 2021; Frantzen et al NDT 2021; Rincon-Arevalo et al, MedRxiv 2021; 8 week data from Goupil et al; Jahn et al Vaccines 2021 and Anand et al MedRxiv 2021 [only the fully vaccinated individuals shown].
Transplant Studies: Sattler et al, MedRxiv 2021; Boyarsky et al, JAMA 2021; Benotmane et al, Kidney Int 2021; Grupper et al Am J Transplant 2021; Benotmane et al Kidney Int 2021; Marinaki et al Am J Transpl 2021; Husain et al KI Reports 2021; Rincon-Arevalo et al, MedRxiv 2021; Chavarot et al Transpl 2021; Yi et al Transpl 2021 and Boyarski et al JAMA 2021.
In the UK, practical solutions that have already been endorsed by the Joint Committee on Vaccination and Immunisation and professional groups include vaccinating eligible adults in transplant households and continuing social distancing, masks etc. Transplant patients should remain cautious and careful in their social contacts and exposures. Vaccinating your transplant waiting list, and your transplant assessment list is prudent. In the UK, the PROTECT-V trial is testing niclosamide as prophylaxis against SARS-CoV-2 infection – and is nationally authorised and adding more sites rapidly – which may prove a useful adjunct whilst we troubleshoot (potentially) lower levels of vaccine protection.
What about the next few weeks or months?
Expect lots more studies! The early data is pretty reassuring for haemodialysis patients, but it cannot be stressed enough that this is a lab surrogate, not proven clinical efficacy, so we should be slow-and-steady in re-normalising dialysis practices.
Transplant remains an open question. The next few weeks of data are going to be very important to work out what to do next – data are coming in from around the world. Possible options include a booster (but of what? mRNA? adenoviral vectors? subunit? Should you hold the antimetabolite too?) and with what antigen (which variant of concern do you need to cover now? and need to cover in 6 months’ time?). The antigen question is likely to need re-visiting for all adults over the next 12 months, so it may be that boosting for HD and transplant becomes the accepted practice to fall in line with the need to cover a new variant.
Guest post for NephJC by Ed Carr,
Postdoctoral Clinical Fellow, Cell Biology of Infection Lab,
Francis Crick Institute, London, UK
Image of an immunoglobulin from Wellcome Images, used under Creative Commons CC BY 4.0 License