Symptom Screens Are Not Sufficient: The Fight Against Tuberculosis Needs Better Weapons

(See the Major Article by Boardman et al on pages 115–20.)

In the United States, 70% of tuberculosis (TB) occurs in non–US-born individuals, making finding and treating TB disease in immigrant populations a critical piece of domestic TB elimination [1]. Screening soon after arrival is particularly important, as the majority of TB disease is clustered in the first few years after immigration [2]. Identifying and treating TB disease among migrant populations can be particularly challenging: mobile populations disperse easily, making TB diagnosis and 6-month-long treatment courses difficult. Additionally, ongoing TB transmission within migrant communities may persist due to poor healthcare access.

In this issue of Clinical Infectious Diseases Boardman et al [3] report on data from one of the most vulnerable groups newly arriving in the United States, individuals being held in detention facilities with US Immigration and Customs Enforcement (ICE). Using chest radiographs as the primary screening tool, they report rates of pulmonary TB disease 30 times greater than seen in the general US population (92.8 per 100 000 compared to 3 per 100 000) [3]. Providers at this facility also performed TB symptom screens with the radiographs, similar to what is used as initial TB evaluation in much of the world (the World Health Organization [WHO] symptoms screen includes cough longer than 2 weeks, hemoptysis, weight loss, fever, or night sweats). Strikingly, 79.2% of individuals with TB disease reported being asymptomatic at diagnosis, revealing the weakness of screening based on symptoms, particularly in high-risk, vulnerable populations. Nearly 20 TB prevalence surveys internationally have shown that over 50% of patients with active TB would be missed with only a symptom screening approach [4]. Additionally, Boardman et al [3] found that being asymptomatic was not associated with clinical signs of being less transmissible, implying lack of symptoms does not mean individuals cannot initiate substantial outbreaks. In contrast, the lack of TB symptoms, or being symptom unaware, could mean individuals would stay in contact with others longer, fueling wider transmission clusters.

The WHO’s End TB strategy targets early identification of TB cases and prompt treatment to interrupt ongoing transmission. However, the tools for initial screening of high-risk groups have not changed in 50 years, still relying on clinical symptoms and, where available, chest radiograph to determine who should proceed to more expensive and technical microbiologic testing. After decades of widely implemented TB screening campaigns across Europe and North America, the WHO recommended in 1974 that “the policy of indiscriminate tuberculosis case-finding by mobile mass radiography should now be abandoned” [4]. This decision was made due to increasing evidence of the mass screening’s inefficiency in populations with low TB prevalence and access to high-quality healthcare [4]. However, striving to reach the elimination goals set first by the Stop TB strategy and now by the End TB strategy, some countries returned to symptom screening in order to close the case-detection gap. In 2013, the WHO published the results of their systematic reviews of TB screening tools, noting “the relative effectiveness and cost effectiveness of screening remain uncertain” [4].

The problem with reliance on symptom screening is 2-fold. First, in some settings, TB stigma may result in underreporting of TB symptoms, and in others, such as ICE detention, fear of consequences of TB diagnosis may lead to symptom denial. Second, there is increasing appreciation of the importance of incipient and subclinical TB, phases when individuals truly do not have symptoms or are symptom unaware [5]. Patients appear to enter a phase where they are at high risk of progression to TB disease (incipient TB) or have culturable Mycobacterium tuberculosis without significant symptoms (subclinical TB). Those with subclinical TB remain potentially infectious, making detection of either of these early stages critical for stopping transmission [6].

Screening with chest radiography, even in targeted populations, has limited specificity due to previous TB lung scarring, recent successfully treated TB, and other infections with overlapping patterns with active TB disease [4, 7]. In the methodology used by Boardman et al, microbiologically unconfirmed patients with no change on repeat chest imaging after 8 weeks of TB therapy were removed from the analysis [3]. This highlights the intensity of follow-up and expertise needed to effectively use chest radiographs as screening and diagnostic tools. Globally, access to chest radiography has improved with rollout of mobile X-ray and computer-assisted interpretations, such as CAD4TB (Delft Imaging Systems, the Netherlands) [7]. Access, however, is still far from universal and inability to differentiate old versus active TB in some cases leads to overtreatment.

After almost 100 years of TB screening based on symptoms and/or imaging, we must concede that use of low-sensitivity or -specificity screening tools is losing the battle against the TB epidemic. What is needed for higher-burden populations, such as US-held detainees, is a more sensitive triage test to detect those who need the more expensive (and more specific) confirmatory diagnostics [8]. No currently available TB triage test meets the validity (≥90% sensitivity, ≥70% specificity) or feasibility criteria (≤US$2 per test, non–sputum-based, available at the point of contact) recommended by the WHO [9].

One approach to increase the sensitivity of TB screening algorithms is to incorporate rapid molecular diagnostics, like Xpert MTB/RIF or Ultra (Cepheid, Sunnyvale, CA). In high-TB-prevalence, low-resource areas this approach has overloaded laboratory systems [10] and misses as many as 25% of TB cases [7]. In low-TB-prevalence, high-resource settings there was hope that the increased sensitivity of Xpert Ultra might lead to earlier TB diagnosis [11]. In these settings, significant false-positive rates are the concern [12]: it is unclear whether lower bacillary range (trace) results represent false positives or subclinical disease. Additional approaches that have shown some promise, although also with limited specificity except in specific populations, include C-reactive protein [13] and lateral flow urine lipoarabinomannan [14].

Several new technologies are showing promise for early detection of TB disease. One is the identification of changes in the host transcriptome associated with active TB. These mRNA signatures have been narrowed from the original 393-gene signature of active TB to a small pool of candidate gene signatures, which would allow near-point of care diagnostics [15, 16]. Furthermore, while transcriptional signatures appear to successfully diagnosis active disease, they may even enable identification of TB before the symptoms start [15]. These signatures appear to detect individuals at high risk of progression 3 to 6 months before they develop disease, meaning testing in detention centers could also capture those at highest risk of later disease [16]. Another avenue for identification of those at high risk of progression to active TB is the use of serum metabolic profiles; such profiles have predicted as early as 12 months prior to the event those at highest risk of progression [17]. Finally, serum protein signatures have also shown promise as potential near-patient diagnostics for active TB [18].

To break the TB transmission chains in migrant populations, we need screening tools to identify TB independently of symptoms. With a better triage test, we could more accurately identify early and even incipient and subclinical TB disease. A successful test should be deployed in settings where immigration from high-TB-burden countries occurs into the United States to meet our domestic goals of TB elimination. These tests should ideally meet the WHO criteria [9], so that they could be used in low-resource settings as well. After decades of the same tools, it is past time for us to upgrade our TB screening armamentarium.

Notes

Financial support. T. C. B. was supported for this work by the National Institutes of Health [T32, DA013911] and the Burroughs Wellcome Fund/American Society for Tropical Medicine and Hygiene Postdoctoral Fellowship in Tropical Infectious Diseases.

Potential conflicts of interest. The authors: No reported conflicts of interest. Both authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.