‘Vaccine resistant’ pertussis – in a way, yes. In a more accurate way, no.

This post is a discussion of the February 2013 report from Philadelphia that the pertussis bacterium may be losing expression of the vaccine target protein pertactin, and the implications of such a loss. A discussion of the April 2014 paper that found most Australian pertussis bacteria have lost pertactin expression can be found here.

A recent letter to the New England Journal of Medicine is making the news rounds online. With the title Pertactin-negative variants of Bordetella pertussis in the United States (full text here) it describes the findings that some of the whooping cough bacteria (Bordetella pertussis) in the US appear to have stopped expressing one of the proteins targeted by the vaccine. With headlines like “Whooping cough may be becoming resistant to vaccines” and “Some Whooping Cough Strains Now Outsmarting Vaccine” circulating in response to the report, it is important we seriously consider the results’ implications when it comes to efficacy of the pertussis vaccine.

Firstly, what do the findings even say? Well, first we need to understand the current pertussis vaccines. Back in the 1940s a pertussis vaccine was introduced that contained whole killed bacteria (a whole cell pertussis vaccine, or ‘wP’ for short). This had a dramatic effect on incidence of disease and associated deaths. However, as the disease declined, the more serious side effects of this vaccine (febrile seizures, dizziness and fainting) became more of a concern, and uptake of the vaccine dropped, leading to a resurgence of disease. In response to this, vaccines were developed with fewer pertussis components – less bacterial content meant less general reactions, a cleaner vaccine all around. These new shots that contain a handful of proteins rather than a whole cell are referred to as acellular pertussis (aP) vaccines.

The currently available aP vaccines in the US and Australia contain either three or five pertussis components (they also always come with diphtheria and tetanus toxoids, and are sometimes also combined with polio, hepatitis B ad Hib vaccines).

Three pertussis proteins are in all aP-containing vaccines – PtxA, Prn and Fha.

PtxA is the active subunit of the pertussis toxin (though it is inactivated for the vaccine). The majority of the disease caused by B. pertussis is caused by the toxin, so the presence of anti-toxin antibodies before infection is a prime goal of this vaccine.

Prn and Fha (short for pertactin and filamentous haemagglutinin [see why I’m using abbreviations?]) are adhesion molecules. In order for pertussis to infect a human it must latch on to our respiratory tract, and that is exactly what these proteins are for. The other two proteins included in only some vaccines (two types of fimbriae – Fim for short) are adhesion molecules too.

Okay, that’s probably enough background, you get it, aP vaccines target the pertussis toxin and either two or four adhesion molecules. Because most aP-containing vaccines (4 of 7 and 5 of 7 brand names in the US and Australia, respectively) have just the three proteins, I’ll discuss the recent results at relevant to immunity induced by these three-component aP shots.

The researchers who wrote the letter in question analysed pertussis bacteria from 12 children hospitalised with the infection in Philadelphia in 2011-2012. They found that of these twelve pertussis isolates, only one had Prn protein. Yet, by the standard test, all of these bacteria would have been considered positive for it.

You see, the protein can vary between individual bacteria, with the variations in pertactin being numbered, Prn1, Prn2, Prn3, etc. The way the specific variation is normally tested for is that the variable part of the gene is sequenced, allowing researchers to categorise the specific Prn-type the isolate is carrying. The researchers found mutations outside of the variable region that rendered the gene unable to produce Prn protein. As such, although all twelve isolates registered as having Prn type 2 by the standard DNA-based test, eleven were negative for the actual protein, something that only DNA sequencing of the whole gene revealed.

Not only does this raise the problem of vaccine evasion (less proteins targeted by the vaccine should hypothetically equate to less protection), Prn-negative mutants are not new, but this raises the possibility that Prn-negative mutants could have been circulating for longer than we think, undetected by the standard Prn-classifying technique.

The fact of the matter is that these are just the results from twelve clinical isolates. All eleven of the Prn-negative strains were otherwise Prn2, so one could naively generalise to the whole country (Prn2 is the predominant type in both the US and Australia). However, the researchers identified three different mutations that caused Prn-negativity in their letter, so rather than being one pertussis type that lost Prn expression and then became the predominant one, it looks like Prn-negativity has cropped up independently multiple times in the already prevalent Prn type 2 variant.

We don’t know the true prevalence of Prn-negative strains, or how long they’ve been around, however I doubt it will be long before we do. Take for example, this study from mid last year. The authors took 661 different pertussis isolates collected in the US from 1935 (before the pertussis vaccine) to 2009, stored by the CDC, and analysed their genomes for a number of different measures of diversity, including typing the proteins targeted by the vaccine, to see how selective pressure from the various vaccines through the years may have affected the wild pertussis populations throughout the country. It would be the work of a few months for the researchers to re-culture the bacteria and test for the mutations that stop Prn expression – possibly only weeks should they have kept the DNA samples produced during the study. I would be fascinated to see how recently these variants emerged, and whether this emergence correlated with any particular change to the vaccines or vaccine schedule – information that could lead to smarter vaccine design.

Okay, so we don’t know how prevalent these strains are, but let’s assume the worst, that all wild B. pertussis populations have lost expression of pertactin. How would that influence vaccine-induced immunity? Well, the current best efficacy estimates of 3-component aP-vaccine induced immunity, as established by the Cochrane Collaboration, is 84-85%. You may have seen some news early last year stating a new strain was evading vaccine-induced immunity thanks to a mismatch in pertactin. There is a meme in the scientific literature that pertactin induces ‘type-specific’ antibodies; that is, antibodies induced by type 1 pertactin in the vaccine don’t bind the Prn2 most wild strains carry. As I discussed in more detail here, this is incorrect, a study on human Prn immunity from 2008 shows most anti-Prn antibodies bind parts of the protein that don’t change between types, meaning that result should have little to no influence on this estimate of vaccine efficacy.

Okay, so what might the efficacy of aP shots be reduced to? Well, the loss of Prn expression effectively reduces three-component aP vaccines to two component vaccines that only target PtxA and Fha. And it just so happens that the Cochrane Collaboration also covered the efficacy of such two component vaccines, with the two studies that met their rigorous inclusion criteria reporting efficacy rates of 59% and 69%.

Okay, that’s a pretty big drop – 84% down to 59-69%, but what do the numbers mean in real terms? I have a real qualm with the standard ‘vaccine effectiveness’ measurement, as it presents the available data in a rather obtuse way which, although no doubt second nature for statisticians, is rather inaccessible for everyone else. The measure is calculated by comparing the percentage of vaccinated in the study that get the disease and the percentage of unvaccinated in the study that get the disease, with 0% efficacy representing an identical rate of disease in both the vaccine-recipient and non-recipient groups (which of course means the vaccine does nothing) and 100% representing complete protection of the vaccinated, but not the unvaccinated. The formula can easily be re-arranged to give the relative frequency at which the unvaccinated catch the disease compared to the vaccinated.

An 84% vaccine efficacy equates to a 6.25x greater rate of pertussis disease in those who did not receive the vaccine compared to those who did not. This contrasts with 59% and 69%, which equate to a 2.4x and 3.2x greater rate of pertussis in controls who did not receive an aP-containing vaccine.

So, in real-world terms, this is clearly a significant decrease. However, it is just as obviously the case that it’s far better to have received the vaccine than not, with a 2.4-3.2 times greater disease rate in the unvaccinated kids.

Okay, so that’s the worst case scenario, but there’s still a few unknowns. Firstly, we simply don’t know how prevalent these Prn-negative strains are (though, like I said above, I imagine that question will be answered soon). Secondly, we don’t know what disadvantage the bug is at by losing pertactin. I said above that Prn is an adhesion molecule, with adhesion being critical to causing disease. As you might imagine, in a mouse model of pertussis infection, losing Prn expression led to a statistically significant decrease in the amount of pertussis bacteria in the lungs and trachea of infected animals, so it’s possible this reduced vaccine efficacy will be mitigated slightly by the reduced colonisation efficiency of these mutants. However, for all the speculations unknowns are still unknowns, and only continued research will change that.

So, what can be done, both by the individual patient, and the vaccine designers? Well, you as a patient can request a vaccine formulation with more pertussis proteins. While from my reading of the literature the role of immunity to the Fim proteins is less than that to PtxA, Fha or Prn, it is still not to be sniffed at. In both Australia and the US it is Sanofi Pasteur whose aP vaccines have five components, with all the GlaxoSmithKline aP shots being three component only. That said, if your doctor only has a three-component aP shot, remember it’s certainly better than nothing! (For those interested, I’ll list the respective brand names at the end of this post – I figure if anyone wants to take this seriously I may as well just list the brands)

What about vaccine designers? Well, the obvious solution is to include more pertussis proteins in the shot. Work into which pertussis proteins elicit an immune response is done, just take this sort of study, in which mice are immunised with wP or infected with live pertussis. The antibodies from these mice are taken, and their target proteins identified as a way of finding the pertussis surface proteins that best elicit an antibody response.

Rather than adding a few surface proteins here or there, what might be better is adding the rest of the cell. Yes, the initial wP vaccines had those mentioned side effects, but as you might imagine, research into better whole cell vaccines has continued into the aP era. Take for example, this phase III clinical trial of a whole-cell pertussis vaccine with reduced side effects. Or this live-attenuated B. pertussis strain, though only just past the first phase of clinical trials.

In the long run, the solution will no doubt be a better pertussis vaccine, and the current research is all incredibly promising, with multiple options coming up in the medium to long term. However, more immediately, the first thing will be for other researchers to see if these same results can be seen in other pertussis isolates throughout the country, and throughout the world. However, even in the worst case scenario for the implications of these results, the vaccine is still effective, and as safe as ever. So get your boosters, the same as you ever would, and just keep an ear out over the next few months as this story evolves, right alongside pertussis.

5 vs. 3 component aP vaccines

In America the 5-component aP vaccines include Daptacel, Pentacel and Adacel; the three component aP vaccines being Infanrix, Kinrix, Pediarix and Boostrix.
In Australia the 5-component aP vaccines are Adacel and Adacel Polio; the three component vaccines are Infanrix-Hexa, Infanrix-IPV, Infanrix-Penta, Boostrix and Boostrix IPV. For what it’s worth, all the brand names of the 5-component vaccines are  registered (?) trademarks of Sanofi Pasteur, the 3-components of GlaxoSmithKline.