Latent MDR-TB (Part 2): the invisible within the invisible
In Part 1 we discussed the nature of latent TB (LTBI), and some of the current (and recurrent) deficiencies in treating it.
In this part, we discuss specifically how LTBI is treated and, in relation to this, how much of the immense pool of latent infection might already be drug-resistant because this is a critical issue.
So how is LTBI treated?
The most recent WHO guidelines include four options for treatment of LTBI, including three new shorter drug regimens
A daily dose of isoniazid for at least 6 months (this often is extended to 9 months).
A weekly dose of rifapentine and isoniazid for 3 months.
A daily dose of rifampicin plus isoniazid daily for 3 months.
A daily dose of 3–4 months of rifampicin.
It’s reckoned that a 60-90% risk reduction of developing active TB is achievable using these treatments. It should be further added that (for reasons that will become obvious below) further regimens are being tested including using second line drugs like levoflaxicin and delaminid, but it’s highly questionable whether either of these options will be affordable at any scale.
The shortcomings of these current recommendations
Rifampicin and isoniazid are the two cornerstone medications in first line TB drug therapy which sounds like a good thing. But, a simultaneous resistance to both of them defines an infection as being multi-drug resistant (MDR-TB), but there also may be mono-resistance to either.
Rifampicin and rifapentine are both from the same group of rifamycin antibiotics, and have unique sterilizing activity against TB but because they come from the same group should generally be considered to have the same characteristics in terms of drug-resistance. Rifapentine is in there because it’s longer acting than rifampicin and therefore has the advantage of being used in a more easily managed weekly dose in treatment of latent TB.
By far the most commonly used treatment is still the first option, however, and this is largely because of its general safety profile. Isoniazid monotherapy has in fact been recommended for some years, but all the three newer shorter regimens are expected to help patients to better adhere to (and complete) their treatment and so may well be used with increasing frequency.
Unfortunately, they all come with similar caveats, relating to a simple question: what happens to treatment outcome (and indeed the patient) if the latent infection is already drug-resistant to any of these drugs (i.e. the patient has acquired it from an infectious contact who carried a drug-resistant strain of TB)?
It’s a reasonable question to ask because, if a latent infection is already MDR, it will de facto be resistant to all four of these approved options. Furthermore, if it’s mono-resistant to rifampicin (and therefore also to rifapentine because they belong to the same drug group) it will be resistant to three treatment options. And if its mono-resistant to isoniazid it will also be resistant to three of them, but particularly to the regimen which is currently the most commonly used.
And of course, if the patient is resistant to any of the drugs used to treat them, then their infection is unlikely to clear, their risk of developing disease will persist and they will have endured at least three months of strong antibiotic chemotherapy for nothing.
So how much latent TB is drug-resistant?
The answer to this is inherently uncertain. This is because, while all diagnostics for latent tuberculosis rely on immune responses (and are fairly reliable), none can determine whether a strain of infection is resistant to any anti-TB drugs. In other words, drug-resistance cannot be clinically identified in latency and this is a massive unresolved handicap to TB control.
Because of this, the issue of invisible drug resistance in latency has been troubling us in Moxafrica for years because, of course, its quantity, if it were measurable could tell us how much MDR-TB is coming down the track, and how much trouble we may or may not be in. Finally, at the start of last month a fascinating paper was published in the Lancet which has addressed it.[i]
The paper’s headline findings were as follows:
19.1 million people may have been latently infected with MDR-TB during 2014 (the baseline year from which data was calculated).
MDR strains of TB are estimated to have accounted for 1.2% of the total prevalent latent global burden in 2014 (or around 20 million cases), and it disproportionately comprised 2.9% of the burden in children under 15.
Recent latent infection with MDR-TB during 2014 meant that 1.9 million individuals were at high risk of developing active MDR-TB in 2015.
These proportions of latent infection that are MDR are increasing and look like they will continue to increase.
Unfortunately, we had several questions to ask about these numbers (and we have raised them with the lead author who unfortunately has yet to reply to our questions).
Our concerns principally focus around the following two issues:
Their estimated percentage of 1.2% of the global latent burden being MDR-TB.
The high probability that the most widely used and well-established preventative therapy (isoniazid preventative therapy or IPT) is encountering a far higher rate of drug-resistance to TB than 1.2% (their estimate for MDR-TB).
The 1.2% proportion of the global burden of latent TB that is MDR
Regarding the first concern, we suspect that there may be an incongruity lurking in their calculation (which we have identified to the authors).
As mentioned, this crucially important paper suggests that 1.2% of all latent infection in 2014 were MDR. We believe that this is a misleading estimate (because of the chronicity of many of the infections and the fact that more recent infections are believed to more readily re-activate), but we also have some concerns about it.
The WHO, for instance, has been estimating for the past 6 years that 3.5-3.6% of all new incident cases of TB are MDR - i.e. nearly 3 times this paper’s estimated latent MDR rate. (We feel obliged to add that we are far from convinced that the WHO percentage is an accurate estimation anyway). But notwithstanding this, since a latent infection is by definition a precursor for an incident case of new re-activated disease, in a stable scenario we believe it logical that the proportion of infections that may be MDR in latency (i.e. in relation to the whole pool of latency) will be something quite similar to the percentage of new incident TB cases that are believed to be MDR (i.e. something around 3.5%, or otherwise the WHO estimate is probably a lot too high).
In fact, the only factor we can think of that might explain why the latent percentage might be less than the incident rate would be if the majority of MDR strains were considered to be ‘fitter’ than the drug-susceptible strains. If this were the case, then MDR strains would have a higher capacity to break out of latency and reactivate in greater proportion than ordinary strains. The problem with this explanation is that the paper itself suggests the opposite: based on limited data (which may be questionable anyway) the paper actually assumes a 40% loss of fitness for MDR strains compared to susceptible ones. (This assumption is certainly not universally accepted, by the way, but the authors may have felt they needed to be cautious).
But if their assumption of loss of fitness is indeed correct, then we reckon that logic dictates that the rate of MDR within the latent pandemic should actually be higher than that of new infections (not lower): for a 3.5% proportion of new cases to be MDR, then we think that the proportion in latency would have to be greater than 3.5% and would need to be in a proportion which could account for and relate to that 40% loss of fitness.
But of course this is illogical as well. If the MDR bacteria really were less fit, then how could a higher percentage of MDR have found its way into the latency pool in the first place given that a percentage of 3.5% exists in the infectious environment that feeds it (unless, of course, DR strains are actually more fit to have enabled this happen)? So this reasoning seems flawed as well.
Actually this may not be quite as illogical as what we have just suggested.
First of all, the WHO’s estimate of the global infectious environment actually comprises a higher percentage of MDR-TB than 3.5% because it also contains the relapse TB cases which are notified as being MDR as well (and 20% of relapse cases are diagnosed as being MDR-TB). To allow for these extra cases it’s been being reported for several years that the overall rate of MDR in all TB cases (i.e. including the relapse ones) is around 5%, and these are also quite probably infectious.
(This 20% rate in relapse cases is probably the most accurate data that we have available, incidentally, but even this can’t be considered that reliable since many relapse cases will never be notified, and many will not be tested for drug susceptibility because of limited diagnostics).
But given that these cases hike the overall proportion of drug-resistance that might be out there and infectious above 3.5%, it certainly suggests that a gradient of percentages across older infections amid latent infections could well exist. This idea is certainly supported by the study's estimate is MDR-TB is nearly three times higher in latent infections in children - since by definition they will have been infected more recently in a period in which more MDR-TB will have been circulating.
But there’s another factor that adds another logical log to the fire that suspects that the rate in the pool of latency might be higher than 1.2% (or more accurately be closer to the percentage of re-activated TB that is MDR). This is simply because more MDR cases generally survive undiagnosed and untreated for much longer periods than DS cases, so they have more opportunity to infect others. While the period before any treatment intervention may be similar, the vast majority of MDR cases never get treated for MDR at all, and can easily live for three or even five years. The majority of drug-susceptible cases (DS-TB), meanwhile, do get treatment. What’s more, the majority of them will no longer be infectious within a couple of months of starting therapy; the same cannot be said for thos MDR-TB cases who do get treatment.
Here’s a quick analysis:-
The current WHO estimates are that around 40% of DS-TB cases generally never get treated at all but, of the 60% of those who do, 82% see successful outcomes. What this amounts to is that overall about 50% of all TB cases are believed to be getting cured (82% of 60%), with the other 50% of them remaining potentially infectious till they either spontaneously cure or die.
With MDR-TB these numbers are very different: 75% of them are reckoned never to see diagnosis or treatment and, of the 25% of cases who do, only 55% see successful treatment outcome. This makes for 13% of MDR-TB getting cured (55% of 25%) and so 87% of MDR cases remain infectious till they either spontaneously cure or die. The odds seem to well stacked to make a case that there is proportionately much more opportunity for MDR strains to be feeding the latency pool than DS-TB ones.
Add to this the associated extra delay in first diagnosing MDR-TB, and this must without question be the case, and we think that this would surely outweigh any loss if fitness even if it were 40%. As such, we would suggest that the idea that 1.2% of the latent pool is MDR should be carefully interpreted. We fear it may be conservative but it is probably not a useful estimate anyway from an epidemiological aspect. Given that the probability of disease reactivating is highest in the first year after infection it's probably the estimate of latent MDR-TB in children that we should be most worried about since this must surely be a reflection of recent infections.
Estimated upwards trends of latent MDR-TB within the overall latent pool
This makes further sense when we note the paper's specific prediction that “the proportion of latent tuberculosis caused by MDR strains will increase”. This conclusion is based on the trends which have already been measured or which they have calculated.
This is important because the trends are far from modest: the authors suggest in a series of graphs (see below) that the percentage of latent TB that is MDR may be roughly doubling every ten years. If this is the case then the overall 1.2% percentage they calculate for 2014 must be nearer 2% today. But if we're right in thinking it's more probable that it was around 3.5% in 2014, however, then it might be nearer 5% today.
What's interesting is that similar trends are not being reflected in WHO reports which have been estimating percentages of incident TB that is MDR so there may be significant issues here for TB control. This is because, if these trends are accurate (and there are certainly many experts who would agree with them), then the balance between existing reactivated MDR-TB, latent MDR-TB and the next wave of reactivated MDR-TB is (unlike for DS-TB) dynamic and far from stable.
(This is 'Figure 2' from the Lancet article). You can see that each graph clearly shows an upward trends of latent MDR-TB in every region. These trends are in stark contrast to what's happening in latent 'all' TB generally which seems to be relatively stable with around 1.8 billion prevalent latent infections and 10 million incident cases each year.
The paper unfortunately doesn’t include graphic or numerical details of any global trend but, given the graphs above, it seems that a rough doubling of the proportion of the latent pandemic that's drug-resistant every decade may not be far from the mark. And if this is the case then it suggests an annual increase of perhaps 7-8% which is not insignificant.
The further probability that isoniazid preventative therapy (or IPT) is encountering a far higher rate of drug-resistance to TB than 1.2%
Of further note, this important paper fails to mention something of relevance. Despite being co-authored by Rein Houben and Peter Dodd who co-wrote the previous important 2016 paper that revised the global 32% prevelance rate of LTBI down to 23%, it makes no mention of their own previous calculation that 10.9% of all latent infections in 2014 were isoniazid resistant. As such, this calculation from three years ago suggests that in 2014 isoniazid mono-resistance in latency may have been nine times more frequent than the current paper’s estimate of MDR for the same year. That's a massive difference which implies that as of today at least 11% of all IPT treatments should be expected to fail (and allowing for their own trends this could even be 15%). What's more. this failure rate will occur whether or not they are MDR which could have huge implications for TB control.
This calculation of 10.9% LTBI being isoniazid resistant is far from unique or extreme, incidentally. A 2015 paper reckoned that 12.1% of children had isoniazid resistant disease,[ii] and if this is the case then the very target group that is in such focus worldwide because of its vulnerability may already be vulnerable to the treatment designated for it, with kids in some countries particularly so. What’s more, a set of South African serial surveys, found an increase in isoniazid resistance from 6.9% to 12.4% in paediatric TB isolates between 1994–2005 (this suggests conformation of the MDT trends suggested by the Lancet paper, but also suggests that it could be more than double this now in these countries).[iii]
Similarly, between 2002 and 2005 in two former Soviet oblasts (Uzbekistan and Azerbaijan) it was reported that isoniazid-resistance in all ages was as high as 40%–49%. And, of course, again it will probably be even higher than that now.[iv]
We believe that these anomalies deserved highlighting in the paper, if not including an actual warning that the most used treatment of LTBI (i.e. IPT) should be urgently rigorously reviewed because it may be out of date because of localised drug-resistance in many countries.
In welcome contrast, rifampicin mono-resistance is reported to be much less frequent (the WHO has estimated that its proportion may be as low as 1.1% of all TB worldwide). What’s more, WHO guidelines for treatment of active disease indicate that it should be treated as if it were MDR-TB anyway and so the same should surely apply to treatment of latent RR-TB infection. If so, however, then this perhaps this 1.1% proportion should be added to any estimate of latent MDR-TB as a precautionary principle. In any case, it leaves all parties in a terrible quandary as to how best to approach treating latent TB generally.
It can be seen that estimating drug-resistance in latency is fraught with uncertainty. But more than this, it’s clear that treating latency is becoming an uncertain matter as well. The current WHO-approved regimens must now be suspect, and new diagnostics and regimens are desperately needed.
In Part 3 (in specific relation to this) we will present a set of recent findings on the use of low dosage daily moxa for treating LTBI that have been harvested in North Korea and discuss their potential implications.
[i] Knight G, McQuaid CF, Dodd PH, Houben RMG. Global burden of multidrug-resistant tuberculosis: trends and estimates based on mathematical modelling. Lancet Infect Dis 2019 http://dx.doi.org/10.1016/s1473.3099(19)30307-x
[ii] Yuen CM, Jenkins HE, Rodriguez CA, Keshavjee S, Becerra MC. Global and Regional Burden of Isoniazid-Resistant Tuberculosis. Pediatrics. 2015;136(1):e50–9.
[iii] Schaaf H, Marais B, Hesse:ling A, et al. Childhood drug-resistant tuberculosis in the Western Cape Province of South Africa. Acta Paediatrica. 2006;95(5):523–528
[iv] Wright A, Zignol M, Van Deun A, et al. Epidemiology of anti-tuberculosis drug resistance 2002–2007: an updated analysis of the Global Project on Anti-Tuberculosis Drug Resistance Surveillance. Lancet. 2009;368(9553):2142–2154.