Empiric tuberculosis treatment and 12-month mortality among sputum GeneXpert-negative adults living with HIV in Uganda in the era of widespread Antiretroviral therapy: A prospective cohort study

This prospective cohort study in Uganda reveals that despite widespread antiretroviral therapy, Xpert-negative adults living with HIV face high 12-month mortality, which is significantly elevated among those receiving empiric tuberculosis treatment, suggesting that negative molecular results should prompt extensive evaluation for alternative diagnoses like cryptococcosis and bacteremia rather than immediate empiric TB therapy.

The Gist

Imagine you are a doctor in a busy hospital in Uganda. You have a patient living with HIV who is very sick, has a bad cough, and a fever. You suspect they have Tuberculosis (TB), a deadly lung infection. You run the best, most modern test available (called a "GeneXpert" or "Xpert"), which is like a high-tech metal detector for TB bacteria.

The test comes back negative. The metal detector didn't find anything.

But here's the problem: In people with weak immune systems (HIV), the bacteria can hide so well that even the best metal detector misses them. So, you are stuck with a difficult choice:

1. Do nothing: Wait and see if they get better, but risk them dying if it is TB.
2. Treat anyway: Give them powerful TB medicine immediately, just in case, even though the test said "no." This is called Empiric Treatment.

This study asked a big question: In the modern era where most HIV patients are on life-saving HIV medication (ART), does giving TB medicine to people with a negative test actually save lives, or does it just add more stress to their bodies?

The Story of the Study

The researchers followed 300 patients in Uganda who fit this exact scenario: HIV positive, very sick, suspected TB, but a negative Xpert test. They watched them for a year to see who survived.

Here is what they found, explained with some simple analogies:

1. The "Sick Room" vs. The "Waiting Room"

The study split patients into two groups: those in the hospital (Inpatients) and those at home (Outpatients).

• The Analogy: Think of the hospital patients as people stuck in a burning building (very sick, low immune counts). The outpatients are people with a small fire in their kitchen.
• The Result: The people in the "burning building" (hospital) were much more likely to die. In fact, 31% of all patients died within a year, and most of those deaths happened in the first three months. Being in the hospital was the biggest predictor of death.

2. The "Shotgun" Approach (Empiric Treatment)

About 23% of the patients (68 people) were given TB medicine immediately, despite the negative test. The doctors were essentially saying, "We can't prove it's TB, but you look so sick, let's shoot first and ask questions later."

• The Surprise: The group that got the TB medicine actually died at a higher rate than the group that didn't.
• The Metaphor: Imagine two groups of hikers lost in a foggy forest.
  • Group A (No TB meds) keeps walking, hoping to find the path.
  • Group B (TB meds) starts taking heavy, toxic pills just in case they are being chased by a bear.
  • The Outcome: Group B died more often. Why? Because the pills didn't cure the real problem. The "bear" wasn't TB; it was something else entirely.

3. The "Imposter" Problems

The researchers looked closely at the patients who died to see what actually killed them. They found that many of these patients didn't have TB at all. They had "Imposter" diseases that looked like TB but were actually something else:

• Cryptococcosis: A fungal infection in the blood (like a hidden mold growing inside).
• Bacteremia: Bacteria in the blood (like a bacterial infection spreading through the veins).
• Cancer: Specifically lymphoma.

The Tragic Mix-up: Many of the patients who died had these "Imposter" diseases. Because the doctors gave them TB medicine (which doesn't kill fungi or bacteria), the real killer kept growing while the patient's body was weakened by unnecessary drugs. It's like trying to put out a grease fire with water—it just makes things worse.

4. The "Safety Net" (ART)

Most of the patients (82%) were already taking their HIV medication (ART). This is like having a strong safety net.

• The Result: Patients on ART survived better than those who weren't. However, even with this safety net, the "sick room" patients (those in the hospital) still had a very high risk of dying.

The Big Takeaway

This study teaches us a hard lesson about medical detective work:

Just because a patient is very sick and looks like they have TB, doesn't mean they do have TB.

When a modern test says "No TB," giving TB medicine immediately might not help. In fact, it might hide the real problem (like a fungal infection or cancer) and delay the correct treatment.

The Solution?
Instead of just guessing and giving TB pills, doctors need to be like detectives with a full toolkit. They need to:

1. Check for other "Imposter" diseases (fungus, bacteria, cancer).
2. Look deeper than just the lungs.
3. Only give TB medicine if they are very sure, or if they have ruled out the other killers first.

In short: In the fight against HIV and TB, "guessing" isn't enough anymore. We need to find the real enemy before we start the war, or we might end up fighting the wrong battle and losing the patient.

Technical Summary

Study Overview

Title: Empiric tuberculosis treatment and 12-month mortality among sputum GeneXpert-negative adults living with HIV in Uganda in the era of widespread Antiretroviral therapy: A prospective cohort study.
Setting: Mulago Referral Hospital (inpatients) and Kisenyi Health Centre-IV (outpatients), Kampala, Uganda.
Study Period: November 2017 – December 2020.
Population: 300 adults living with HIV (PLHIV) who were clinically suspected of having active tuberculosis (TB) but tested negative on the sputum Xpert MTB/RIF assay.

---

1. Problem Statement

• Diagnostic Limitations: Despite the introduction of the Xpert MTB/RIF assay, a significant proportion of PLHIV with active TB remain undiagnosed due to the test's limited sensitivity in immunocompromised individuals (paucibacillary disease).
• Clinical Dilemma: In high-burden settings like Uganda, clinicians frequently resort to empiric TB treatment (initiating therapy without microbiological confirmation) for Xpert-negative patients based on clinical suspicion to prevent mortality.
• Knowledge Gap: While randomized controlled trials (e.g., REMEMBER, STATIS) have shown no mortality benefit for empiric TB treatment in ART-naïve patients, there is a lack of "real-world" evidence regarding outcomes in populations where the majority are already on Antiretroviral Therapy (ART). It remains unclear if empiric treatment improves survival in Xpert-negative PLHIV who are on ART, or if it masks other fatal comorbidities.

2. Methodology

• Design: Prospective cohort study.
• Inclusion Criteria: PLHIV aged ≥18 years, clinically suspected of active TB (per WHO guidelines), and confirmed Xpert-negative.
• Exclusion Criteria: Patients already initiated on anti-TB medication prior to enrollment.
• Data Collection:
  • Clinical: Demographics, TB symptoms, Karnofsky Performance Score (KPS), Chest X-ray (CXR), and ART status.
  • Laboratory: CD4 count, Serum Cryptococcal Antigen (CrAg), Urine TB-Lipoarabinomannan (TB-LAM), Sputum Mycobacterial Growth Indicator Tube (MGIT) culture, and Blood cultures.
  • Follow-up: Monthly telephone follow-ups for 12 months to ascertain vital status.
• Primary Outcome: 12-month all-cause mortality, stratified by empiric TB treatment status.
• Statistical Analysis: Kaplan-Meier survival analysis, log-rank tests, and Cox proportional hazards/Poisson regression models to calculate mortality rates and adjusted mortality rate ratios (aMRR).

3. Key Contributions

• Real-World Evidence: Provides data from a routine clinical setting in a high-burden country where ART coverage is high, moving beyond the controlled conditions of previous RCTs.
• Comprehensive Etiology Search: Unlike many observational studies, this research actively investigated non-TB etiologies (cryptococcosis, bacteremia, other mycobacteria) in a cohort of Xpert-negative patients to determine if missed diagnoses contributed to mortality.
• Differentiation of Settings: Explicitly compares outcomes between inpatients and outpatients, highlighting the disparity in disease severity and mortality risk.

4. Key Results

• Demographics: Median age 37 years; 55.7% female; 82.3% on ART; Median CD4 count 206 cells/mm³.
• Empiric Treatment: 22.7% (68/300) received empiric TB treatment. These patients were significantly sicker: lower CD4 counts (median 87 vs. 373), lower BMI, lower KPS, and more likely to be inpatients (77.9% of treated group).
• Mortality:
  • Overall: 31.0% (93/300) died within 12 months.
  • Timing: 72% of deaths occurred within the first 3 months.
  • Setting: Inpatient mortality was 91.4% (of deaths), compared to 72% of deaths occurring in outpatients (though absolute numbers were lower for outpatients).
• Impact of Empiric Treatment:
  • Unadjusted: Patients receiving empiric TB treatment had significantly higher mortality rates (51.5 per 1,000 person-months) compared to those not treated (30.2 per 1,000 person-months; p=0.013).
  • Adjusted: After adjusting for confounders (care setting, functional status, ART), empiric treatment was not an independent predictor of mortality (aMRR 1.05; p=0.850). The high mortality in the treated group was driven by the severity of their underlying illness (low CD4, inpatient status, poor functional status).
• Alternative Diagnoses (Non-TB Etiologies):
  • TB Culture: 5.0% (15/300) were culture-positive despite Xpert negativity.
  • Cryptococcosis: 12.3% had positive serum CrAg.
  • Bacteremia: 3.7% had positive blood cultures.
  • Missed Diagnoses: Among the 93 deaths, 17.2% had positive CrAg and 7.5% had positive TB cultures. Notably, many of these patients were not treated for the specific identified pathogen (e.g., cryptococcal treatment was not documented).

5. Significance and Conclusion

• High Mortality persists: Even with widespread ART, Xpert-negative PLHIV face high mortality, particularly in the first 3 months and among hospitalized patients.
• No Survival Benefit from Empiric TB: The study confirms that empiric TB treatment does not improve survival in this population. The observed higher mortality in the treated group was due to confounding by indication (clinicians treated the sickest patients).
• Diagnostic Pitfalls: A significant portion of deaths were associated with undiagnosed or untreated non-TB conditions, specifically cryptococcal meningitis and bacteremia.
• Clinical Implications:
  • Clinicians should avoid reflexively prescribing empiric TB treatment for Xpert-negative PLHIV without ruling out other fatal opportunistic infections.
  • Expanded Diagnostics: There is an urgent need to integrate routine testing for Cryptococcus (CrAg), blood cultures, and TB-LAM/culture in Xpert-negative PLHIV, especially those with advanced immunosuppression (low CD4) or inpatient status.
  • Focus on Comorbidities: Management should prioritize investigating and treating non-TB etiologies rather than assuming TB is the cause of death in Xpert-negative cases.

Final Takeaway: In the era of widespread ART, empiric TB treatment for Xpert-negative PLHIV does not reduce mortality. The high death rate is likely driven by missed diagnoses of other opportunistic infections (Cryptococcus, bacteremia) and the severity of advanced HIV disease, necessitating a broader diagnostic approach beyond TB-specific testing.