The Value of Lateral Flow Urine Lipoarabinomannan Assay and Empirical Treatment in the Xpert MTB/RIF Ultra Era: a Prospective Cohort Study


 Introduction: The value of Lateral Flow urine Lipoarabinomannan (LF-LAM) assay and the role of empiric tuberculosis (TB) treatment in the era of the highly sensitive Xpert MTB/RIF Ultra (Xpert Ultra) assay is unclear. This study aimed to assess the additional yield of Xpert Ultra compared to Xpert MTB/RIF, the yield of LF-LAM assay when used in combination with Xpert Ultra, and the role of empiric TB treatment in the Xpert Ultra era among patients with symptoms of TB admitted to a tertiary hospital in Ethiopia. Methods: We performed a secondary analysis of a cohort of hospitalized Xpert MTB/RIF-negative patients. Sputum samples were examined for Mycobacterium tuberculosis by culture and Xpert Ultra. In HIV positive and severely ill patients, a urine sample was collected for the LF-LAM assay. Treatment outcome was assessed six months after enrollment. Logistic regression was used to identify factors predictive of deaths among Xpert Ultra negative patients. Results: Of the 250 hospitalized Xpert MTB/RIF-negative patients, 35 (14.0%) had culture-confirmed TB. The Xpert Ultra assay diagnosed 31 of the 35 culture positive cases that had been missed by Xpert MTB/RIF. The LF-LAM assay did not identify any case that was not detected by Xpert Ultra. Empiric treatment did not improve survival among Xpert Ultra negative patients (OR 1.28, 95% CI: 0.3-5.3). Low body mass index (< 18.5kg/m2) was the only significant predictor of death in Xpert Ultra negative patients (OR 4.0, 95% CI: 1.12-16.5).Conclusions: In this prospective cohort, we observed that LF-LAM did not improve the diagnostic yield when used in combination with Xpert Ultra and did not find a survival benefit of empiric treatment among Xpert Ultra negative individuals. Research is needed to determine the optimal management of patients with presumptive TB and a negative Xpert Ultra assay in high TB burden settings.


Introduction
Tuberculosis (TB) is a chronic infectious disease caused by Mycobacterium tuberculosis (MTB) complex [1]. With an estimated 10 million active TB cases, of which 10% are co-infected with HIV, and 1.3 million deaths in 2019, TB remains a public health problem worldwide [2]. In 2019, Ethiopia had the 10th highest incidence of TB globally, with 114, 233 new cases and an incidence rate of 151 cases per 100, 000 people [2,3].
Even though the World Health Organization (WHO) endorsed the molecular Xpert MTB/RIF assay almost a decade ago [4], many primary care clinics in high burden countries still rely on smear microscopy as the initial diagnostic test. Because of the low sensitivity of smear microscopy, smear negative patients with TB symptoms are often empirically started on TB treatment [5]. Even when the Xpert MTB/RIF assay is available, empirical treatment remains highly prevalent in high TB burden countries [6]. The use of empirical treatment can be justi ed in these settings given the suboptimal performance of the Xpert assay, especially among smear negative individuals (sensitivity of 67%) and in people living with HIV (sensitivity of 80%) [7,8].
In 2017, the WHO recommended replacing Xpert MTB/RIF by Xpert MTB/RIF Ultra (Xpert Ultra) as the initial diagnostic test because of its superior performance in smear negative patients and patients living with HIV [9]. A meta-analysis of 19 studies found an overall pooled sensitivity of 84% the Xpert Ultra compared to 69% for Xpert MTB/RIF [10]. In smear-negative TB cases ,Xpert Ultra was 17% more sensitive compared to Xpert MTB/RIF; in people living with HIV Xpert Ultra was 13% more sensitive [11][12][13]. Whether empiric TB treatment is still justi ed in high TB burden settings in the Xpert Ultra era has not yet been investigated.
The lateral ow urine lipoarabinomannan assay (LF-LAM) is another tuberculosis diagnostic tool that was recently endorsed but is not yet commonly used in high TB burden countries [14]. In inpatient settings, LF-LAM can assist in the diagnosis of TB in people living with HIV and in people who are seriously ill. Among people living with HIV who present with signs and symptoms of TB to an inpatient setting, the pooled sensitivity of LF-LAM was 52% (40-64%) and pooled speci city 87% (78-93%). When used in combination with other tests, LF-LAM can improve the diagnostic yield by 14% (compared to Xpert MTB/RIF alone) or 36.6% (compared to clinical sign and symptoms alone) [15,16]. It has also been shown that the use of LF-LAM can improve the survival of patients hospitalized with advanced HIV disease [pooled risk ratio for mortality 0.85 (0.76-0.94)] [14,17,18]. The yield of LAM when used in combination with Xpert Ultra has not yet been assessed.
In this study, we aimed to assess the additional yield of Xpert Ultra compared to Xpert MTB/RIF, the yield of LF-LAM assay when used in combination with Xpert Ultra, and the role of empiric TB treatment in the Xpert Ultra era among patients with symptoms of TB admitted to a tertiary hospital in Ethiopia.

Study setting
The study was conducted in Jimma University Medical Center, a tertiary hospital located in Jimma, Oromia Region, Ethiopia. Patients are referred to the Center from health centers, district hospitals and private health facilities in Jimma Zone and district hospitals in neighboring regions, resulting in a catchment population of over 20 million. The hospital has about 800 inpatient beds and about 20 000 inpatients, 220 000 outpatient, and 15 000 emergency cases visit the hospital annually [19]. During the study period (December 2018 to July 2019), the Xpert MTB/RIF assay was used as the initial test for diagnosing TB in all people presenting with signs and symptoms of TB at the Jimma University Medical Center [20]. The Mycobacteriology Research Center of Jimma University, which is located in close proximity of the medical center, serves as the TB reference laboratory for southwest Ethiopia.

Study population and data collection
This analysis is a secondary analysis of a cohort study that aimed to determine the role of empiric treatment among hospitalized adults (age ≥ 18 years) with symptoms of pulmonary TB (current cough, night sweats, fever, and weight loss) who tested negative on the Xpert MTB/RIF assay [21]. The cohort study enrolled 125 patients treated empirically for active TB and 125 patients in whom the physician decided not to treat for TB. A structured questionnaire was administered by trained nurses to collect demographic and clinical characteristic and medical records of the patients were reviewed to document HIV status, CD4 count, and improvement of TB symptoms on TB treatment. Sputum samples were collected in the rst 48 hours following enrolment for a single Xpert Ultra assay, liquid culture using the Mycobacteria Growth Indicator Tube (MGIT) system and solid culture on Lowenstein Jensen (LJ). Drug susceptibility testing for rst line anti-TB drugs (rifampicin and isoniazid) was done by Line Probe assay (LPA). HIV-positive and severely ill patients (temperature>39 o C, respiratory rate >30 resp./min, cardiac rate >120bpm, or unable to walk without help) [22] were asked to provide a 10 ml urine sample for a LF-LAM assay. All laboratory tests were performed at the Mycobacteriology Research Center of Jimma University.

Laboratory procedures
Mycobacterial culture Sputum sample decontamination and digestion, inoculation of concentrated sputum samples on LJ and MGIT media were performed using standard procedures [23]. The mycobacterial culture was classi ed as positive when growth was detected by LJ and/or MGIT with a positive SD Bioline TB antigen MPT64 con rmatory test and a negative blood agar test. Culture negative denotes the absence of growth on both LJ and MGIT culture or positive LJ/MGIT culture with a negative SD Bioline TB antigen MPT64 result.
Xpert MTB/RIF Ultra test Xpert Ultra test was performed according to manufacturer's instruction [24]. Brie y, 1 ml of sputum was added to 2 ml of reagent in a 15-ml Falcon tube, vortexed for at least 10 seconds and incubated at room temperature for 10 minutes. The mixture was then added to the Xpert Ultra assay cartridge and placed in the instrument. The Xpert Ultra test was repeated using the same sample in cases of an invalid result. In cases of "trace" result, a new sputum sample was collected for a second Xpert Ultra test. If the second test was positive (trace or higher), the result was considered positive for patients living with HIV and for HIV-negative patients without a history of TB treatment.

LF-LAM test
The urine Determine TB-LAM test (Abbott Laboratories, Lake Bluff, USA; formerly Alere Inc, Waltham, USA) was performed by applying 60μL of fresh urine to the sample pad at the bottom of the test strip using a micro-pipette. After 25 minutes of incubation at room temperature, the test strip was inspected visually. If any band is observed, the intensity was scored as grade 1, 2, 3 or 4 as compared with the reference card. The test was read independently by two certi ed laboratory professionals in blinded manner (blinded to each other and the sputum Xpert Ultra and culture results). LAM test results were classi ed as positive if grade ≥ 1 and as negative if no test band was observed [16].

Data analysis
Descriptive statistics were performed using frequencies and percentages to describe the characteristics of the study population. The yield of Xpert Ultra and LAM tests was determined by comparison to mycobacterial culture results. Mortality in the 6 months post assessment was assessed among Xpert Ultra negative participants, strati ed by empiric treatment status. To identify factors associated with mortality among Xpert Ultra negative patients, we calculated the odds ration using a logistic regression model. Data analysis was performed using R Statistical software version 3.6.1.

Ethical consideration
Ethical clearance was obtained from the Ethical Review Board of Institute of Health of Jimma University (Ref. no: IHRPGD/397/2018). Written informed consent was obtained from all study participants. All methods were performed in accordance with the relevant guidelines and regulations. Clinicians treating study participants in whom TB was bacteriologically con rmed were noti ed of the result.

Characteristics of the study population
All 250 patients admitted to the hospital with signs or symptoms of TB who were enrolled in the parent study were included in the secondary analysis. By design, all study participants had a negative Xpert MTB/RIF result; 125 were empirically initiated on TB treatment by the treating physician and the physician did not start TB treatment in the other 125 participants (Fig. 1). The majority were 40 years of age or younger (144/250, 57.6%), female (138/250, 55.2%), and rural residents (154/250, 57.6%). Almost all patients (232/250, 92.8%) had already visited another health facilities for similar complaints before their current hospitalization. HIV status was documented for 243 (97.2%) participants of which 21.4% were positive (Table 1). All participants had symptoms of current cough, fever, night sweat and shortness of breath; 165 (66%) had a cough for ≥ 14 days, 45 (18%) had a history of previous TB treatment and (93/250, 37.6%) had a chest X-ray ndings typical or compatible with TB.  because of an error result on the initial test ( Table 2). The Xpert Ultra assay was positive in 4 patients with a contaminated culture. One rifampicin resistant strain identi ed by LPA was not identi ed as rifampicin resistant by Xpert Ultra.

Yield of Xpert Ultra and role of LF-LAM tests
Participants whose sputum cultures were contaminated (n = 11) or positive for NTM (n = 2) were excluded from this analysis. Of the remaining 237 participants, 35 (14.8%) had culture-con rmed TB. The Xpert Ultra assay detected 31 (88.6%) of the 35 culture-con rmed cases. The 4 culture-con rmed cases that were missed by Xpert Ultra were HIV negative. The assay was positive in 4 (2.0%) and was negative in 198 (98.0%) of the 202 culture-negative participants. The 4 Xpert Ultra positive culture negative cases had very low mycobacterial burden as demonstrated by a high cycle threshold (Ct) value (Ct value of the lowest probe ≥ 29.2, Supplemental Table), were all HIV negative, and none of these cases had a history of prior TB treatment.
The 68 people living with HIV or seriously ill individual assessed by LF-LAM, one participant was excluded from the analysis because of presence of NTM (n = 1). Among the remaining 67 participants, nine LF-LAM tests were positive. Seven of the 11 (63.6%) culture positive cases were detected by LF-LAM. These seven cases were also were positive Xpert Ultra. In addition, two positive LF-LAM assays occurred in four of the Xpert Ultra positive but culture negative participants (Fig. 2). The LF-LAM assay thus did not increase the diagnostic yield when used in combination with Xpert Ultra.

Role of empiric TB treatment in the Xpert Ultra era
Of the 215 (86%) Xpert Ultra negative participants, 94 (43.7%) were diagnosed clinically and empirically started on TB treatment, and 121 (56.3%) were clinically considered not to have TB and were not initiated on TB treatment. Of the 94 Xpert Ultra negative participants started on empiric TB treatment, 90 (95.7%) were alive and 4 (4.3%) had died in the 6 months following assessment. Of the 121 Xpert Ultra negative patients not started on TB treatment, 115 (95.0%) were alive and 6 (5.0%) had died in the 6 months following assessment (Table 3). Among Xpert Ultra negative patients, the odds of survival was similar between those who did and did not receive empiric TB treatment (OR 1.28, 95% CI 0.3-5.3).  Predictor of 6-month mortality among Xpert Ultra negative patients In regression analysis, it was found that underweight or body mass index (BMI) < 18.5kg/m2 was signi cant predictor of mortality among Xpert Ultra negative patients (OR 4.0, 95% CI: 1.12-16.5) ( Table 4).

Discussion
In this prospective cohort study, we con rmed the high sensitivity of Xpert Ultra among cases of presumptive TB, as 89% (31 of the 35) of the culture positive cases missed by Xpert MTB/RIF were positive on Xpert Ultra. We further demonstrated that the use of empirical treatment may no longer be justi ed in the Xpert Ultra era and that the recommendation for use of a LF-LAM may need to be revised for settings where access to Xpert Ultra is available.
These observed high sensitivity of Xpert Ultra is in line with a meta-analyses, where 87.2% of TB culture positive patients were positive on Xpert Ultra test [12] and an overall estimated pooled sensitivity of 90.9% (95% credible interval 86.2 to 94.7) [25]. Similar to other studies, we also observed potentially false positive Xpert Ultra results, as 4 cases were Xpert Ultra positive by culture negative. False positive Xpert Ultra results have been associated with a recent history of TB treatment [11]. In our study, the Xpert Ultra positive culture negative cases had no history of previous TB treatment. Interestingly, two of the four cases had a positive LF-LAM result, suggesting that these cases may actually be true positive instead of false positive cases of TB.
The LF-LAM assay detected 63.6% of the culture positive cases among people living with HIV or seriously ill presumptive TB cases. This corresponds to ndings of a study performed in four countries in sub-Saharan Africa where a 60.0% sensitivity of LF-LAM was observed among culture positive cases [26]. In contrast to studies performed during the Xpert MTB/RIF era, where addition of LF-LAM to Xpert MTB/RIF assay increased the yield, the LF-LAM assay did not improve diagnostic yield as all LF-LAM positive patients were also positive on Xpert Ultra. These suggesting a limited role for LF-LAM in the Xpert Ultra era and, if con rmed in a future, larger study, may require revision of the guidelines for use of LF-LAM. In hospital settings where a Xpert Ultra assay is available, all LF-LAM negative cases would have to be reassessed by Xpert Ultra due to poor LAM sensitivity, and all LF-LAM positive cases would also need to reassessed by Xpert Ultra to determine whether there is resistance to rifampicin.
In high TB burden settings, treatment decisions often continue to rely on clinician's judgement and chest radiography ndings, even when it has been shown that empiric treatment does not result in survival bene t [6,27]. In this study, we found that the Xpert Ultra assay detected almost all culture positive cases, questioning the value of reliance on empiric TB treatment for patients with negative Xpert Ultra results.
Furthermore, we found that empiric treatment of Xpert Ultra patients did not increase the odds of survival, as survival was similar between those who did and did not receive empiric TB suggest that prescribing empiric treatment should be re-assessed as this may pose an unnecessary burden the health care system and may expose patients to unnecessary treatment. Future research should investigate which alternative diagnoses should be considered in these patient in order to develop evidence-based guidelines for the management of Xpert Ultra negative patients.
Our study had some limitations. First, we limited enrollment to hospitalized Xpert-negative patients, which limits generalizability of the outpatient settings. Second, our study was not powered to assess factors associated with mortality among people with presumptive TB who had a negative Xpert Ultra assay.
Lastly, we did not assess the presence of other respiratory pathogens to explore the cause of symptoms or death in the Xpert Ultra negative patients.
In conclusion, Xpert Ultra assay provides a sensitive, speci c and rapid diagnosis of TB among presumptive pulmonary TB cases. The use of LF-LAM test did not yield additional cases in Xpert Ultra negative patients. The odds of survival was similar between Xpert Ultra negative patients who did and did not receive empiric TB treatment. Our ndings question the value of empiric treatment and the LF-LAM assay in the Xpert Ultra era. Future studies should be performed to provide the evidence needed to establish guidelines for the management of Xpert Ultra negative patients.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.