Rapid CRP as screening test for TB in PLHIV

By Madhukar Pai Moderator | 03 Sep, 2017

Please see these two new papers on the potential of rapid C-reactive protein (CRP) to screen individuals for active TB among people with HIV. One is a meta-analysis, while the other is a primary study from Uganda.

It will be great to have a discussion on the strengths and limitations of CRP as a triage test in TB.





Sandeep Saluja Replied at 9:14 PM, 3 Sep 2017

Interesting!Would need to try it out on a few of my patients.

Ruth MCNERNEY Replied at 7:23 AM, 4 Sep 2017

Hi Madu

To kick off the discussion: I think the major problem with CRP is the specificity which appears to vary quite widely in different patient populations. This may mean it is more useful in some settings and patient groups than others, which will make life difficult for policy makers deciding where and when such a test should be used. However, I don't think we shall know the true specificity of CRP until we have data on extra-pulmonary TB.

It would be interesting when testing patients with low CD4 counts to see comparative data from the Alere LAM TB test which is a low cost, rapid POC test that would seem to have a lower sensitivity but a higher specificity than CRP.

I think we need studies that compare all the 'easily measurable' potential TB biomarkers in a series of well defined patient populations. Looking at biomarkers/tests one by one will take too long if we are really going to eradicate TB in the foreseeable future. We should also tell the funding bodies not to fund studies to evaluate blood/urine based POC tests unless they include rigorous investigation for extra-pulmonary disease.

best regards

Claudia Denkinger Replied at 1:09 PM, 4 Sep 2017

Thank you for sharing these papers, Madhu. I think in the absence of a TB-specific triage test, this could certainly be explored further across different clinical setting. The test should be optimized though to achieve high sensitivity in order not to miss patients with TB that can be transmitted. Or it should be explored in an algorithms with symptom screening.
It is promising to see that the specificity is holding up (>60%) in HIV-positive patients across CD4 strata with certainly a high variety of co-morbidities. The data on non-HIV patients is sparse and I would like to see further data in areas where there is more malaria and helminth co-infections. Ideally, as Ruth points out these studies would have a more rigorous gold standard that captures more EPTB. This might explain some of the losses in specificity.

Kavitha Saravu Replied at 1:54 PM, 4 Sep 2017


Interesting discussion.

It would be worthwhile to test the performance of CRP as a screening test in low HIV, high TB burden settings, as one of the limitations of the meta analysis was that 4/5 outpatient studies were performed in South Africa.

Agree with the idea that such studies should have rigorous investigations for EPTB, which would also identify alternate diagnoses eg malignancy, alternate chronic infection which is true in India.

In one of our studies, we found 1/3rd of presumptive TB subjects had malignancy and another 1/3rd had other chronic infections among HIV negative subjects.(in press)

Heidi Hopkins Replied at 2:28 PM, 6 Sep 2017

Apologies for joining this discussion late - in reply in particular to Claudia's notes re: "The data on non-HIV patients is sparse and I would like to see further data in areas where there is more malaria and helminth co-infections. ..."

This will not address the TB concerns head-on, but in case it's of interest:
We are preparing a study of fever etiologies and antimicrobial resistance across five sites in Africa and SE Asia that vary in HIV and malaria prevalence as well as in epidemiology of other infectious diseases. In addition to rigorous testing for specific fever-causing pathogens, we will test for a panel of host biomarkers of immune/inflammatory response including CRP.

We are intentionally excluding patients with prominent respiratory or gastrointestinal infections, so are likely to miss pulmonary TB; but may encounter a fair amount of EPTB. At the moment we have resources only to culture blood for mycobacteria (in addition to other assays for several other bacteria, viruses and parasites) - but if anyone would like to discuss alternative and/or novel diagnostics for EPTB we are certainly open to ideas and further collaboration.

We plan to enroll 2400 febrile patients (inpatients and outpatients, children and adults) plus 600 community controls per site, for a total of 12,000 patients and 3000 controls across the five sites.

Happy to discuss further if useful,

Vijay Nema Replied at 12:12 AM, 7 Sep 2017

Dear Dr. Hopkins
Good to know about your work. I am proposing a similar study but at a lower scale in India. Specifically teasing out the unknown fever etiologies and antimicrobial resistance. I would like to learn from your experience on this. Please share your protocols and guide me if feasible.
Thanks and Best regards

Christina Yoon Replied at 2:02 AM, 13 Sep 2017

Hi all,
Thanks to Madhu for initiating this discussion.

I'm one of the authors of the recent CRP pubs. I agree with Ruth's and others comments that CRP specificity (and sensitivity) can be expected to vary by population. In fact, the specificity of any test will vary by prevalence of non-TB disease that also causes CRP elevations (e.g., bacterial infections). For example, if you were to screen hospitalized patients for active TB using CRP, specificity will be poor because such patients are sick with TB or non-TB disease that required them to be hospitalized. Similarly, sensitivity of CRP as well as any other test can be expected to vary by the spectrum of TB severity in the population. Using the same above example, CRP can be expected to have higher sensitivity and higher median levels in an inpatient setting as compared to an outpatient setting since hospitalized TB patients are likely to have more advanced TB than ambulatory TB patients. Based on these principles of test accuracy, we hypothesized that if applied to a general population of outpatient PLHIV initiating ART, CRP would have sufficiently high specificity (and sensitivity) for active TB. I'm really looking forward to seeing if future studies will also find CRP to be a useful TB screening tool in other populations.

Regarding EPTB: I'd be surprised if any single test were to have the same sensitivity/specificity profile in PTB and all types of EPTB. Some forms of EPTB cause minimal systemic symptoms/inflammation (i.e., TB lymphadenitis) while other forms of EPTB are associated with profound inflammation (i.e., TB peritonitis). Just as sensitivity of Xpert MTB/RIF varies with respect to specimen type, I'd expect sensitivity of all tests including CRP, to differ based on type of EPTB. Future studies planning to study the diagnostic accuracy of CRP (or any test) for EPTB should probably look at specific types of EPTB rather than lumping them all together.

FYI: we considered investigating patients in our study for EPTB but quickly realized that this would be a huge undertaking (would require a substantially larger sample size to ensure adequate numbers of EPTB types and extensive testing to classify EPTB status and type). These realities and our interpretation of the current guidelines (screening is intended to increase PTB case detection for better PTB outcomes and decreased TB transmission) led us to focus on PTB.

Also, I just wanted to address Claudia's comment about specificity and malaria burden. Uganda (our enrollment site) is regarded as a country with high malaria burden. Unless severe, parasitic and helminth infections shouldn't cause elevations in CRP which is stimulated by predominantly pyogenic infections causing elevations in IL-6 but who knows, I haven't found a whole lot published on this. I'd expect that specificity of CRP would be at least similar if not better in settings with less malaria burden than Uganda.

Great comments/discussion - I hope it continues.

Isra Cruz Replied at 2:54 AM, 13 Sep 2017

Hi, thanks for this very interesting discussion.
Just to keep in mind re specificity of CRP testing in specific settings: high CRP levels are also common in active visceral leishmaniasis (kala-azar).
All the ebst