Diagnostic accuracy of the Fujifilm SILVAMP TB LAM II (FujiLAM II) assay for the diagnosis of childhood tuberculosis

Abstract

Background

Point-of-care tests (POCTs) are urgently needed to improve tuberculosis (TB) diagnosis in children. This study evaluated the diagnostic accuracy of the Fujifilm SILVAMP TB LAM II (FujiLAM II) assay for childhood TB diagnosis.

Methods

Spot urine samples were collected from children with TB (either bacteriologically confirmed or clinically diagnosed) and from healthy controls, then tested using FujiLAM II. Sensitivity was assessed against microbiological and clinical reference standards, while specificity was determined against healthy controls.

Results

Of 137 participants, 87 had TB and 50 were controls. FujiLAM II showed overall sensitivity of 17.2%, higher in bacteriologically confirmed cases (36.8%) than in clinically diagnosed cases (11.8%). Sensitivity was better in children under five (27.9%), HIV-positive children (25.0%), and underweight or stunted children (24.0%). Specificity was 100%. In the subgroup of 66 children tested with both assays, detection rates did not differ significantly (28.8% vs. 22.7%). When results from either test were considered positive, the detection rate increased to 40.9%, representing a 12.1% improvement over Xpert Ultra alone.

Conclusion

FujiLAM II shows suboptimal sensitivity for childhood TB diagnosis, with better sensitivity in children under five years, HIV-positive, or malnourished, and may add value to current diagnostic algorithms.

Background

In 2024, an estimated 1.2 million children developed tuberculosis (TB), yet only 664,000 (55.3%) were officially reported, highlighting a major diagnostic gap [1]. This gap stems from the inadequacy of current diagnostics, which rely mainly on sputum. Most children cannot produce sputum, and even when they can, the paucibacillary nature of TB lowers diagnostic sensitivity [2, 3]. For children unable to produce sputum, alternatives such as induced sputum, gastric aspirate, or nasopharyngeal swabs are invasive, unpleasant, and yield poor accuracy. Capacity to perform these procedures is also often lacking in high TB-burden settings [2–5]. The World Health Organization (WHO) recommends molecular detection of M. tuberculosis in stool as a non-invasive diagnostic alternative [6]. However, stool-based Xpert testing shows variable performance and faces practical challenges, since not all children can provide a stool sample on demand [7, 8]. Moreover, stool-based molecular testing has not yet been fully rolled out in most countries [2]. As an accurate and accessible diagnostic test for TB in children is still lacking, clinicians often rely on chest radiographs and clinical signs. This approach is complicated by non-specific signs and symptoms that overlap with common childhood illnesses such as pneumonia and malnutrition [9, 10].

Urine is a promising sample for childhood TB diagnosis due to easy access and non-invasive collection. Currently, the AlereLAM test, which detects the mycobacterial cell wall component lipoarabinomannan (LAM), is the only WHO-approved urine-based point-of-care test (POCT). However, its use is restricted to severely ill HIV-positive patients due to suboptimal sensitivity [11]. The Fujifilm SILVAMP TB LAM (FujiLAM, Fujifilm, Tokyo, Japan), a next-generation LAM test developed to address this limitation, utilizes high-affinity monoclonal antibodies targeting M. tuberculosis-specific LAM epitopes combined with a silver amplification step to enhance assay performance, and has shown better accuracy than AlereLAM [12]. In children, the first-generation FujiLAM showed variable sensitivity, ranging from 11% to 65%, with specificities between 83.3% and 95% [13–15]. This version had lot-dependent variability, making it unsuitable for broad clinical use [16, 17]. The manufacturer has since modified the assay and introduced Fujifilm SILVAMP TB LAM II (FujiLAM II), which reportedly resolves this variability [18]. Nevertheless, its diagnostic accuracy across different patient groups, including children, remains largely unevaluated. Therefore, we aimed to evaluate the diagnostic accuracy of FujiLAM II for diagnosing TB in children, a population that stands to benefit the most from a urine-based point-of-care assay.

Methods

Study population and design

Children aged ≤ 14 years were recruited in Addis Ababa, Ethiopia, between October 2023 and August 2025. This study was a prospective cross-sectional study and included a TB group and a control group. The TB group was recruited from five hospitals, namely Zewditu Memorial Hospital, ALERT Hospital, St. Peter Specialized Hospital, Yekatit 12 Hospital Medical College, and St. Paul’s Hospital Millennium Medical College, while controls were recruited from a primary school, namely Bruh Tesfa Primary School. Participants were enrolled consecutively during the study period. The TB group included children with bacteriologically confirmed or clinically diagnosed TB. Confirmed cases were defined as those with a positive Xpert MTB/RIF Ultra (Xpert Ultra, Cepheid, Sunnyvale, CA, USA) result from sputum or gastric aspirate. Clinically diagnosed TB cases were defined as children who were Xpert Ultra-negative or untested and for whom a decision to initiate TB treatment was made by a physician based on clinical assessment, suggestive chest radiograph (CXR) findings, known TB exposure, or lack of response to broad-spectrum antibiotics. Children who had received more than three days of anti-TB treatment prior to enrollment were excluded. Attending physicians who made the diagnosis were blinded to the FujiLAM assay results. The control group consisted of apparently healthy children with no signs, symptoms, or known history of TB.

Sample collection, processing, and transport

Spontaneous spot sputum samples were collected from children able to expectorate. For those unable to produce sputum, gastric aspirate samples were obtained by attending physicians as part of routine clinical care. Briefly, depending on the child’s age, after a fasting period of 4–6 h, approximately 20 mL of sterile saline was instilled via a nasogastric tube, and the aspirate was collected. Both gastric aspirate and sputum samples were tested using Xpert Ultra at the study sites, following the manufacturer’s instructions as part of routine clinical diagnostic procedures. Spot urine samples were collected in sterile urine cups; for younger children, urine was collected after diaper removal by waiting for spontaneous voiding directly into the cup. Urine samples were transported from the study sites to the Armauer Hansen Research Institute (AHRI) under cold chain conditions within 2 h of collection. At AHRI, samples were aliquoted and stored at − 80 °C, although a subset was tested fresh.

Clinical and demographic data collection

Structured questionnaires were used to collect demographic and clinical data, completed through family interviews and medical record reviews. Nutritional status was assessed using WHO Anthro version 3.2.2 for children under 5 years and Anthro Plus version 1.0.4 for those aged ≥ 5 years [19, 20]. Nutritional status was classified using z-scores as follows: underweight was defined as a weight-for-age z-score < -2 (or BMI-for-age z-score < -2 for children > 5 years), and stunting as a height-for-age z-score < -2.

FujiLAM II analysis

During the early phase of enrolment, the FujiLAM assay was temporarily unavailable; therefore, urine samples collected during this period were stored at − 80 °C until testing could be performed. Stored urine samples were then thawed at room temperature prior to analysis, while freshly collected samples were processed directly; all samples were vortexed to ensure homogeneity before testing. The FujiLAM II assay was performed according to the manufacturer’s instructions. Briefly, 200 µL of urine was pipetted into the reaction tube, vortexed, and incubated at room temperature for 40 min to allow antigen–antibody interaction prior to application to the test device. The mixture was then remixed, applied to the test strip, where the complexes were subsequently captured by immobilized antibodies, and processed by activating button 2 until an orange indicator appeared, followed by button 3. Results were read within 10 min after control band development. A test was considered positive if both control and test bands were visible (including faint lines), negative if only the control band was present, and invalid if the control band was absent. All tests were read independently by two readers, one of whom was blinded to the reference standard results. Invalid or discrepant results were repeated once.

Statistical analysis

Data were summarized using descriptive statistics. Categorical variables were compared using the χ² or Fisher’s exact test. The sensitivity of the FujiLAM II assay, with 95% confidence intervals (CIs), was calculated against three reference standards: Xpert Ultra, clinical diagnosis, and a composite standard. Specificity was determined using healthy controls. The diagnostic performance of FujiLAM II was compared with Xpert Ultra using McNemar’s test. Associations between FujiLAM II positivity and clinical subgroups (HIV status, nutritional status, age) were assessed by multivariable logistic regression, with results reported as adjusted odds ratio (aOR) and 95% CIs. All analyses were conducted in R (version 2024.12.0) via RStudio, with p value < 0.05 considered significant.

Ethics approval and consent to participate

This research was conducted in accordance with the principles of the Declaration of Helsinki. The study protocol was reviewed and approved by the ALERT/AHRI Ethical Review Board (approval number: PO-48-23). Prior to enrollment, written informed consent was obtained from the parents or guardians of all participating children. Assent was sought from children aged above 7 years.

Results

Demographic and clinical characteristics of the study participants

A total of 137 children were enrolled: 50 healthy controls (36.5%) and 87 with TB (63.5%). The TB group was significantly younger than controls (median 5 years [IQR 1.6–8.5] vs. 10 years [IQR 9–11], p < 0.001). Gender distribution did not differ, with 45 (51.7%) females in the TB group and 22 (44.0%) in controls (p = 0.488). Detailed demographic and clinical characteristics are shown in Table 1. Among the 87 children with TB, 19 (21.8%) were bacteriologically confirmed and 68 (78.2%) clinically diagnosed. The distribution of TB types among the 87 children was as follows: pulmonary TB (PTB) in 47 cases (54.0%), extrapulmonary TB (EPTB) in 23 cases (26.4%), and disseminated TB in 17 cases (19.5%).

Table 1.

Demographic and clinical characteristics of the study participants (n = [137])

Parameter		All Study Participants (n = [137])	TB		Healthy controls (n = [50])	P-value
			Microbiologically Confirmed TB (n = [19])	Clinically Diagnosed TB (n = [68])
Age groups	0–4	42 (30.7)	10 (52.6)	32 (47.1)	0 (0.0)	< 0.001
	5–9	52 (38.0)	4 (21.1)	24 (35.3)	24 (48)
	10–14	43 (31.4)	5 (26.3)	12 (17.6)	26 (52)
Gender	Male	70 (51.1)	9 (47.4)	33 (48.5)	28 (56)	0.681
BCG scar	Yes	78 (56.9)	9 (47.4)	39 (57.4)	30 (60)	0.137
Underweight	Yes	36 (26.3)	9 (47.4)	23 (33.8)	4 (8)	< 0.001
Stunting	Yes	34 (24.8)	7 (36.8)	24 (35.5)	3 (6)	< 0.001
TB categories	PTB	47 (34.3)	15 (78.9)	32 (47.1)		0.003
	EPTB	23 (16.8)	0 (0.0)	23 (33.3)	NA
	Disseminated TB	17 (12.4)	4 (21.1)	13 (19.1)
TB contact	Yes	12 (8.76)	7 (36.8)	5 (7.35)	NA	< 0.001
TB history	Yes	3 (2.19)	2 (10.5)	1 (1.47)	NA	0.051
Cough of any duration	Yes	60 (43.8)	16(84.2)	44 (64.7)	NA	< 0.001
Night sweating	Yes	53 (38.7)	18 (94.7)	35 (51.5)	NA	< 0.001
Fever	Yes	62 (45.3)	15 (78.9)	47 (69.1)	NA	< 0.001
Appetite loss	Yes	65 (47.4)	18 (94.7)	47 (69.1)	NA	< 0.001
CXR Suggestive of TB	Yes	53 (38.7)	16 (84.2)	37 (54.4)	NA	0.075
HIV	Positive	8 (5.84)	1 (5.26)	7 (10.3)	NA	0.676
FujiLAM II	Positive	15 (17.2)	7 (36.8)	8 (11.8)	0 (0.0)	< 0.05

Note: Comparisons for TB contact, TB history, and symptoms of TB were conducted only within the TB group, as we purposely enrolled children who did not have these traits as controls. Additionally, CXR and HIV data were collected only for the TB group. BCG: Bacillus Calmette-Guérin; CXR: Chest X-ray. HIV: Human Immunodeficiency Virus; N/A: Not Applicable; TB: Tuberculosis

Diagnostic accuracy of FujiLAM II for childhood TB diagnosis

Of 137 patients tested with the urine FujiLAM II assay, 135 yielded valid results. Two invalid results, both from the TB group, remained invalid after repeat testing and were counted as negative. Reader disagreement occurred in two cases and was resolved by repeat testing. All control samples were tested fresh. In the TB group, 17 of 87 samples were tested fresh, while the remaining samples were stored for 3 to 10 months prior to analysis.

The overall sensitivity of the FujiLAM II assay was 17.2% (15/87; 95% CI: 10.7–26.5). In univariate analysis, sensitivity was significantly higher in bacteriologically confirmed TB cases (36.8%; 7/19; 95% CI: 19.1–58.9; p = 0.017) than in clinically diagnosed cases (11.8%; 8/68; 95% CI: 6.1–21.5). By disease manifestation, sensitivity was numerically highest for disseminated TB (29.4%; 5/17; 95% CI: 13.3–53.1), followed by PTB (19.1%; 9/47; 95% CI: 10.2–32.9), and EPTB (4.3%; 1/23; 95% CI: 0.8–21.0). Sensitivity was also higher in HIV-positive children (25.0%; 2/8; 95% CI: 7.1–59.1) than HIV-negative children (17.1%; 13/76; 95% CI: 10.3–27.1), and in underweight/stunted children (24.0%; 12/50; 95% CI: 14.3–37.4) compared to those with normal growth (8.8%; 3/34; 95% CI: 3.0–22.9); however, none of these subgroup differences were statistically significant (p > 0.05 for all). The FujiLAM II assay showed significantly higher sensitivity in children < 5 years (28.6%; 12/42; 95% CI: 16.7–44.1; p = 0.014) than in those ≥ 5 years (6.7%; 3/45; 95% CI: 2.3–17.9). In multivariable analysis, after adjusting for HIV status, nutritional status, and disease manifestation, sensitivity in children < 5 years remained statistically significant (aOR = 6.34; 95% CI: 1.64–33.45; p = 0.014). When sensitivity was adjusted for age, neither disease type nor child nutritional status was independently associated with FujiLAM II positivity (p > 0.05). The assay demonstrated 100% specificity, with none of the 50 controls testing positive (50/50; 95% CI: 93.0–100%).

In the subgroup of 66 participants tested by both Xpert Ultra and FujiLAM II, no significant difference in positive rates was observed. Xpert Ultra detected 19/66 positives (28.8%; 95% CI: 18.1–41.5%), vs. 15/66 (22.7%; 95% CI: 13.2–34.7%) for FujiLAM II (p = 0.502). Notably, 68.4% (13/19) of Xpert Ultra positives were from gastric aspirates samples. When positivity was defined as either assay being positive, the detection rate increased to 27/66 (40.9%; 95% CI: 29.1–53.7%). This represents a 12.1% gain over Xpert Ultra alone, with FujiLAM II contributing eight additional positives missed by Xpert Ultra.

Among 76 children with TB who underwent CXR, there was no statistically significant association between radiographic findings suggestive of TB and FujiLAM II positivity (p = 0.216). Positivity was higher in children with normal CXR (30.4%, 7/23) than those with abnormal CXR (15.1%, 8/53). For the 64 children who underwent all three tests (CXR, Xpert Ultra, and FujiLAM II), the overlap between CXR findings suggestive of TB, Xpert Ultra positivity, and FujiLAM II positivity is shown in Fig. 1.

Fig. 1.

Agreement between diagnostic methods. Venn diagram showing the overlap of positive results from chest X-ray (CXR), Xpert Ultra, and FujiLAM II among the 64 children tested by all three methods

Discussion

There is an urgent need to improve TB diagnosis in children through POCT that are simple to use, child-friendly, and utilize non-invasive samples such as urine. In this study, we evaluated a new version, the FujiLAM II assay, for childhood TB diagnosis. The assay demonstrated an overall sensitivity of 17.2% (36.8% in bacteriologically confirmed cases and 11.8% in clinically diagnosed cases), with a specificity of 100%. Previous studies of the first-generation FujiLAM reported variable performance, with sensitivities ranging from 11% to 65% and specificities from 83.3% to 95% [13–15]. These discrepancies may be attributed to differences in patient populations, including factors such as HIV prevalence, nutritional status, and disease severity, as well as variations between FujiLAM kit lots. Unlike the first-generation assay, which exhibited lot-dependent performance [16, 17], the FujiLAM II used in this study is reported to have improved consistency [18]. Therefore, our results likely reflect the true diagnostic sensitivity of the FujiLAM II assay for childhood TB in our study setting.

In this study, although the differences were not statistically significant, the FujiLAM II assay demonstrated numerically higher sensitivity in children with disseminated TB (29.4%) compared with those with PTB (19.1%) or EPTB (4.3%). Similarly, sensitivity was numerically higher in children living with HIV (25.0%) compared to HIV-negative children (17.1%), and in children who were underweight or stunted (24.0%) compared to those with normal growth (8.8%). These findings are consistent with observations from the first-generation FujiLAM assay, which also showed trends toward improved sensitivity in these groups [13, 14]. The higher sensitivity in these groups is likely attributable to greater immunosuppression, higher mycobacterial load, and disseminated disease, which lead to an increased urinary LAM concentration [13, 14, 21]. Notably, FujiLAM II exhibited a statistically significantly higher sensitivity in the younger age group (< 5 years), with a 28.6% detection rate versus 6.7% in children ≥ 5 years. This finding should be interpreted with caution, as potential sources of contamination may have influenced the observed higher sensitivity in younger children. Nonetheless, improved sensitivity in this age group is particularly valuable, as the diagnostic gap is greatest among younger children, who face the highest risk of severe TB progression and mortality [10, 22]. In fact, TB is a common cause of death in children under five [22]. However, mortality can be drastically reduced with prompt diagnosis and treatment, as evidenced by a pooled case fatality ratio of just 0.9% among treated children [22, 23]. A rapid POCT such as FujiLAM II may therefore help facilitate the timely diagnosis needed to initiate treatment in this vulnerable population.

Among children tested with both assays (n = 66), detection rates for Xpert Ultra and FujiLAM II were not significantly different (34.0% for Xpert Ultra vs. 26.4% for FujiLAM II). Importantly, the majority of children (68%) who tested positive with Xpert Ultra in this study provided a gastric aspirate specimen, which is invasive, unpleasant, and requires trained personnel for collection [4]. In contrast, FujiLAM offers a non-invasive, child-friendly, and rapid point-of-care alternative that can be performed by any health worker, including lay health workers [24]. The Xpert platform presents operational challenges such as device malfunctions, maintenance issues, and power dependencies [25, 26], whereas FujiLAM provides significant operational advantages as it requires no electricity or instrumentation, can be used in remote clinics, and delivers results in approximately one hour. This may add value in improving TB detection in low-resource settings.

Our results show that FujiLAM II was not sensitive enough to be used as a standalone test to diagnose TB in children. However, it identified eight additional children who were missed by Xpert Ultra, representing a 12% increase in detection. Furthermore, FujiLAM II positivity did not correlate with CXR findings suggestive of TB. This suggests FujiLAM may detect cases not apparent on CXR and not detected by Xpert Ultra, indicating the tests provide complementary value. As no single test is accurate enough to diagnose TB in children, combining different tests has been suggested to improve overall detection [3]. Including FujiLAM II in such an approach might therefore benefit pediatric TB diagnosis. However, the additional economic burden of incorporating FujiLAM into the current diagnostic algorithm requires further investigation to determine if its advantages outweigh the costs, particularly in resource-limited settings.

In this study, FujiLAM II demonstrated 100% specificity. Previous studies using the first version of FujiLAM reported specificity ranging from 83.3% to 95% [13–15], with lower specificity observed in younger children, likely due to risks of urine contamination during collection [14, 15]. The high specificity in our study may be attributed to the use of control participants who were apparently healthy children aged 6–14 years, for whom clean-catch urine collection is easier. As a result, the 100% specificity reported in this study may represent an overestimation and may not fully reflect specificity in routine clinical settings, particularly in younger children where the risk of urine contamination is higher. Further studies are needed to evaluate FujiLAM II specificity across different age groups, especially in younger children.

The findings of this study should be interpreted in light of its limitations. First, we enrolled apparently healthy children aged 6–14 years as controls; the control group was not fully age-comparable to the TB group. Including symptomatic controls of similar age would have been more appropriate for better assessment of true specificity. Second, we relied solely on Xpert Ultra as the microbiological reference standard without performing culture. Moreover, not all children were tested using Xpert Ultra due to the unavailability of biological specimens, which may have influenced the proportion of bacteriologically confirmed cases. Third, the majority of urine samples from the TB group were stored, which could potentially reduce the concentration of LAM and affect assay performance. However, existing evidence from first-generation FujiLAM studies indicates minimal variation between fresh and stored samples [27]. Fourth, as only one of the readers was blinded, this may have introduced some degree of interpretation bias. Finally, the modest sample size may limit the generalizability of the results.

Conclusion

The findings of this study highlight FujiLAM II’s suitability as a child-friendly POCT with added value in addressing current TB diagnostic challenges in pediatric population. FujiLAM II demonstrated better sensitivity in children living with HIV, those with undernutrition, and children under five years of age. In settings where Xpert Ultra is available, FujiLAM II could serve as an adjunct test to improve bacteriological confirmation, as it identified additional children who were missed by Xpert Ultra.

Abbreviations

AHRI

Armauer Hansen Research Institute

aOR

adjusted odds ratio

CIs

Confidence intervals

CXR

Chest radiograph

EPTB

Extrapulmonary tuberculosis

FujiLAM II

Fujifilm SILVAMP II TB LAM

LAM

Lipoarabinomannan

POCTs

Point-of-care tests

PTB

Pulmonary tuberculosis

TB

Tuberculosis

WHO

World Health Organization

Xpert Ultra

Xpert MTB/RIF Ultra

Author contributions

S.A, A.P and A.M conceptualized the study and secured grant funding. S.A, A.P, T.W, B.W, and A.M contributed to study design, data interpretation, and supervision. T.A, A.B, and W.T performed participant diagnosis and enrollment. S.A, S.B, and B.T handled sample collection and analysis. S.A drafted the initial manuscript. All authors reviewed and approved the final submitted version.

Funding

This work was supported by the Armauer Hansen Research Institute and the NIH Fogarty International Center Global Infectious Diseases grant (D43TW009127).

Data availability

The datasets generated and analyzed in this study are included in this article. Further inquiries can be directed to the corresponding author.

Declarations

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.