Diagnostics and active case finding for TB: what are more sensitive tests actually detecting?

The current HATIP article is the first in a two-part series on developing lab capacity – to support more active TB case finding. Active case finding is important to more quickly identify and diagnose cases of TB that might not otherwise have been diagnosed, and reduce the opportunity for these cases to spread TB in the communities.

There has been growing disagreement among clinicians and TB experts regarding the use of symptom screening versus more TB sensitive diagnostics such as liquid culture and GeneXpert for active case finding (as opposed to routine diagnosis of symptomatic patients).

Symptom screens may miss some cases of TB, and the many people who screen positive eventually need to be referred for the diagnostic process at the laboratory anyway. Liquid culture, which has become established as the gold standard for TB diagnoses, is a marked improvement over solid culture systems in settings that can perform it properly. It can also detect much smaller amounts of MTB. In an active case finding context, such as household contact tracing in high burden settings, specimens from household contacts of a TB patient sometimes become positive on liquid culture even though they are without symptoms. Generally in close, but not perfect, agreement with liquid culture, GeneXpert can also come back positive in contacts without symptoms. But in settings where transmission of MTB is a dynamic process, how should a positive GeneXpert or liquid culture be interpreted in smear-negative cases without symptoms?

The most common interpretation has been that these are cases of subclinical TB that previously could not be detected – and that TB symptom screens might miss. However, there are alternative explanations for what these ‘TB’ cases might be. For instance, some could be recent infections by MTB that, in immune competent hosts at least, are likely to become latent.

We decided to open the question up to our advisory panel and TB experts: 'how do we know 
if the signs of mycobacterial infection that are being picked up by liquid culture or GeneXpert – in an active case
 finding context always represent 
reactivation disease (TB), rather than transient primary MTB infection or reinfection that may yet become latent; or in the case of GeneXpert – merely the genetic debris from dead microbes following a recent infection that has already gone into latency), or the result of poor infection control where the sputum has been collected leading to a contaminated specimen?

You are asking a key question that the adult TB world have chosen to disregard in the race to improve case detection.

Clearly we should improve case detection and a positive culture in a highly immune compromised person is a significant result that probably justifies TB treatment, irrespective of the presence of clinical symptoms or signs of disease. I did acknowledge this in the AJRCCM letter, since this is justifiable from a risk:benefit perspective.

The fundamental question remains though – what is it that we are measuring?

From the natural history of disease studies in children we know that transient M.tb excretion following primary infection is a common entity and that the majority of these children never progress to active disease.

It is reasonable to expect that transient organism excretion may occur following recent re-infection events as well and in certain TB endemic areas in SA, infection/re-infection may occur multiple times each year, which implies that at any one time there may be many people out there - transiently excreting minute amounts of M.tb (that can be detected by very sensitive tests such as culture or PCR) – but likely on their way towards self-containment.

This has been a major issue in pediatric TB case definition discussions – see manuscripts included.

It would also require a lot more critical thinking now that sensitive detection tools are rolled out and used for active case finding. In principle I would caution against “active case finding” amongst completely asymptomatic individuals (as we do in children – where inclusion of asymptomatic kids with “culture confirmed TB” has complicated interpretation of vaccine protective effect and potentially derailed vaccine development progress), unless this is reserved for severely immune compromised pts.

Positive culture/Xpert in an asymptomatic individual - could mean:

- Subclinical or incipient disease (likely in severely immune compromised; may result from primary or re-infection or re-activation)

- Transient excretion following recent primary infection or re-infection (more likely in a high-burden setting; low risk of disease progression in immune competent)

- Transient excretion as a result of subclinical re-activation (more likely in a low-burden setting; not sure how to quantify risk – high in immune compromised as above)

Sorry for brief and possibly contentious comments, but hope this encourages more robust debate.

Ben

Ps. See also radiological classification of child TB where difficulty of transient features in asymptomatic children is discussed.

Ben, Theo and colleagues,

We have followed up asymptomatic (neg WHO symptom screen) HIV+ patients with sputum Mtb-positive cultures in our cohort. Progression to overt symptomatic disease within days to weeks is observed in a majority. This was also observed by Oni et al in another recently published study which is referenced in the attached article. I have no problem treating such culture+ patients (albeit in the context of a reliable lab) and I have not seen a convincing false+ Xpert TB diagnosis and so would readily treat all Xpert+ patients too as this comes to largely replace culture in SA.

But would you make the same decision about asymptomatic smear-negative HIV-NEGATIVE individuals, because those are the people who are being picked up in some of the household case finding studies.

Also, I'd wonder about whether CD4 cell counts matter. I believe Oni's cohort had higher CD4 cells and she observed progression within around 56% (off the top of my head, I'd have to check the figure). And I don't believe
IPT was used by many in that population (might it not work on its own in 
these 'sub-clinical' cases?) I would think one probably should offer full treatment probably to anyone with under 350 CD4 cells and some lab evidence of TB (culture/geneXpert).

But I am not so sure of what should be done in people with higher CD4 cells. I am thinking primarily about people who come from home-based HIV testing or more aggressive HIV testing campaigns. The implication of some of these studies, like the one in Kenya, would be that each of these people should have laboratory screening, rather than regular and routine symptom screening.

I think this would be a VERY expensive strategy and I'm not certain that we have proof that it is absolutely necessary... that existing policies (if employed rigourously and if we don't lose people to follow-up) of symptom screening, and IPT aren't adequate at higher CD4 cell counts. What do people think about this?

Thanks Theo, Let me enter the interesting debate.

From the resource-poor setting of a country like Malawi, I could never see the resources (human, material and financial) being enough to screen for TB in HIV-infected asymptomatic individuals – be this screening with smears, Xpert, urine LAM or culture. To do this equitably and expansively in HIV-infected asymptomatic persons is just too big a job, so the question of finding significant amounts of TB in asymptomatic persons for me is really a non-issue in the real world of poor countries.

We have to focus on identifying TB in PLHIV with symptoms (Getahun’s screening tool) and investigate those with symptoms with the technology the country has decided it can afford. Smear and CXR are inadequate, and if Xpert could be made more point of care and cheaper this would be a major advance. However, even investigating those with symptoms is a massive job, because if I read Getahun’s PLOS Med paper correctly, nearly half of PLHIV screened will have symptoms dictating the need for investigation.

Where to do this? Most obvious place is the ART clinic and, if in existence, the pre-ART clinic. Nice to think we could do this in the community but again outside of the research setting, do we have the resources?

I think Steve Lawn has brought this up, but let me bring it up again. We do active TB case finding so we can identify TB patients so we can treat them so we can cure them. The bottom line is for me is cure, and I think we need to seriously start linking the outputs of active TB case finding with treatment and treatment success. From the limited literature on this, I think we do not do so well. A significant proportion of those TB patients detected by active case finding never start treatment! And this brings me back full circle to the asymptomatic patient – how well do these culture+ve asymptomatic persons do in terms of compliance and adherence with 6-months TB treatment and do we obtain good cure rates?

Thank you Theo and to all who contributed very interesting and important insights.

Clearly there remain many questions to be answered.

Just 2 things to mention that may require consideration:

1) When talking about “asymptomatic patients” then we should not refer to those without “old WHO TB suspect criteria”, since reported coughing for 3 weeks or more etc. clearly has poor sensitivity. The emphasis should rather be on ANY current symptoms; in children we enquire about: cough, fever, weight loss (or failure to thrive) and excessive fatigue (or reduced playfulness) in addition to screening for obvious signs of possible TB such as visible neck nodes.

A very simple and pragmatic strategy to improve specificity (if feasible) is to simply review all symptomatic pts in 1-2 weeks (unless symptoms are highly convincing) – if symptoms persist (despite treatment of the most likely alternative cause) then investigate further. In children at least, most of the transient infections caused by viral infections etc. resolve within this time period – realize people may have concerns about loss to FU etc.

2) Another thing that is rarely discussed is how frequently X-pert/culture-based screening of “asymptomatic” individuals should be performed. Should this be done annually in TB endemic settings where risk of infection is high? With symptom-based screening the guidance would be to screen at every single clinic visit, with further investigation as clinically indicated (eg. with high risk features or any persistent symptoms on clinical FU). This is what we proposed in the TB/HIV review attached (contains additional relevant references).

Realize there are no easy answers and that the correlation with child TB issues is limited, but just wanted to mention these two things.

Ben

Thanks to all for interesting comments

We need more data to inform these debates. Please have your own groups and encourage your colleagues to submit data to the International AIDS Conference coming up this summer – abstract deadline Feb 15. HIV/ TB will be a key topic at this meeting. We have the opportunity to have some terrific discussions there and advance these issues.

Thank you Diane!

I can't emphasize how important it will be for HIV/TB to achieve prominence at the IAC. There will be -- needless to say -- plenty of competition among frenzied, exhausted, jet-lagged attendees for other scientific topics; domestic HIV/TB is not a big concern in the US since Ken Castro's department at CDC, and the states, have the problem much better in hand than, say, the UK -- despite chronic shortages of 2nd line drugs and their exorbitant prices -- but as always the need to educate and inspire the larger AIDS community about HIV/TB remains a huge and unmet one. The last few years have seen an accumulation of exciting data based on the work of many among you, among others, and it really needs to be highlighted in Washington, DC, in July. So please draft and submit those abstracts!

Hi Theo and colleagues for this interesting debate.

Culture-positive asymptomatic TB exists. This has repeatedly demonstrated in high quality studies. What is less clear is whether the proportion of TB patients that are asymptomatic differs by geographic region, and what the risk factors are for asymptomatic TB. Theo and

Ben also raised many relevant issues regarding our incomplete understanding of the distribution of possible causal pathways and the natural history among these individuals.

Most of the data presented in the draft HATIP article is based on presentation of preliminary data. As such, in my opinion, a very important paragraph in Theo's article is "Frankly, to use preliminary results to broadcast to National Health programmes in resource limited settings that TB symptom screens are inadequate and that the only way to detect (all) TB cases responsibly is through the use of expensive diagnostics that everyone wants but cannot afford ... may, at the very least be poorly considered since it will discourage programmes from implementing what they can do to save people's lives.

Just at the time when we have determined a pragmatic approach to exclude active TB, we seem to be willing to "kill" this approach by highlighting the fact that it fails to find "all" TB cases (even the original Getahun analysis did not promise a 100% NPV). From a public health perspective, we need to define the risk benefit ratio of not implementing ICF/IPT using the 4 symptom screen vs implementing ICF and IPT in asymptomatic HIV-infected individuals and consequently inadvertedly placing some people with detectable replicating bacteria on ITP.

Unfortunately, we do not yet have the population based cohort studies to address this issue.

Another issue that may be counterproductive is a quote by Cavanaugh "Screening algorithms should be validated in the region before they are to be rolled out" ... because ... " reporting of symptoms can be influenced by regional and cultural factors". Again, do we really want to give the message that it is fine to postpone the implementation of ICF/IPT until the screening algorithms have been validated in all regions (and how do we define a region that is culturally homogenous in regards to reporting of symptoms?).

Hi Theo --- just seconding Annelies's points

Yes we do know that asymptomatic or subclinical TB exists, but we have a poor grasp of

a) how "stable" this state is (seems like not in most people --- see below).

b) how important it is with respect to transmission --- given that coughing is an important part of spreading infection

Follow-up of our asymptomatic culture+ participants in prevalence surveys (for the research goal of "case-ascertainment" as well as for making sure that disease is treated) has shown that cases who remained cultrue positive had developed sysmptoms by the time a culture result was available --- of course this could well be different with GXP. Sometimes we found that there were symptoms originally that the person didn't want to own up. Sometime the person was now culture-negative and completely fine with no evidence of TB disease over several months of follow-up --- perhaps the "natural history" of recent TB infection that Ben has reviewed. But most such instances in HIV-infected people progress to obvious symptomatic disease within a few weeks.

We published some work from Zimbabwe about 10 yrs ago in TSRU proceedings that suggested that most people progressed rapidly to treatment-seeking for symptomatic disease once they were smear-positive, but a small percentage do remain s+ and undiagnosed for very long periods even when access to culture-based diagnosis was right on their doorstep.

If either resolution to latent TB or rapid progression to symptomatic state is the general rule for almost all asymptomatic cases, then repeated symptom screening (as in HIV clinics) will pick the people who are progressing the next time they attend, and (dare I say it) isoniazid PT will sort out the newly latent ones!

Agree fully with Annelies that there is need to be pragmatic about TB screening guidelines, and that 100% accuracy is not currently a realistic goal. There may be a few parts of the world (ie South Africa!) where the risk/prevalence of subclinical disease is high enough to make screening of asymptomatics worthwhile, but in most parts of Africa and elsewhere globally the costs of detecting this type of disease will blow a huge hole in the NTP budget for little gain and at a time when GF is financially shaky. Also what about culture-negative disease (which will also tend to be GXP negative)? This could be up to 20% of adult cases and higher for kids.

Also, even if you do detect them, the very small number of people who are really in a stable asymptomatic c+ state are not all that straightforward to manage once you do make the diagnosis --- understandably they are reluctant to accept a diagnosis of TB, and need confirmatory tests and the time to talk them into being convinced that they do indeed have TB. This is beyond what can be expected from basic clinic staff in most African settings --- and at worst can go completely the wrong way and generate a lot of anger and being threatened with public health legislation etc.

So for both public health and individual benefit of early diagnosis --- I would also argue that we focus first on implementing TB symptom screening well. Where funds permit and prevalence rates of asymptomatic TB are high, then there can be more comprehensive TB screening regardless of symptoms, but this will not be a realistic goal in most HIV clinics. And in that case there needs to be some special consideration of how you confirm the need for TB treatment in asymptomatic people --- bearing in mind that screening is just that, and that the guidelines for managing screening programmes stress very strongly that a positive screening test should not be assumed to be confirmatory, no matter how good the diagnostics.

Visit the HATIP Archive.