Tuberculosis is an infection caused by MTB bacteria that is most commonly manifested in the lungs.²⁷ Some of the risk factors for TB are age, smoking behavior, having a poor sanitary environment and low socioeconomic factors.²⁸ In this study, 56.4% of pulmonary TB suspect patients were adults (19-59 years) followed by 41% elderly (>60 years). The average age was 54 years, with a higher proportion of male than female (61.5% vs. 38.5%, respectively). However, the results of the chi-square correlation analysis of age and gender showed no statistically significant correlation between age (p = 0.527) and gender (p = 0.196) with the pulmonary TB incidens. One study states that in adults, post-primary tuberculosis is common due to endogenous reactivation or exogenous reinfection in a previously sensitized patient who has maintained some previously sensitized patients, who have maintained some level of immunity.²⁹ Further, several factors are thought to play a role in the mechanism of increased susceptibility in male such as smoking, which can cause lung damage and decreased immune system function, and alcohol use which has an immunosuppressive effect.³⁰

Although HIV and DM are not correlated with TB incidence in this study (p = 0.123 and
p = 0.225), it was found that 75% (3/4) of patients with HIV, and 57.1% (4/7) of diabetes patients had pulmonary TB. These two comorbidities are often cited as influencing the incidence of TB, suggesting that diabetes suppresses the immune response, which in turn facilitates MTB infection and/or progression to symptomatic disease. In addition, poor glycemic control adversely affects TB treatment outcomes with effects such as prolongation of culture conversion, treatment failure, relapse and death.³¹ Other study states CD4⁺ cells attacked by HIV causing the number of CD4⁺ T cells decrease, so that the remaining T cells work excessively, resulting in exhausted T cells, where the peak of the decrease in IFN-g and TNF-a results in increased MTB activity.³²

In the description of clinical symptoms in patients with suspected pulmonary TB such as cough, shortness of breath, weight loss, fever, night sweats and hemoptysis, does not correlate with the incidence of pulmonary TB (p = 0.123, p = 0.151, p = 0.520, p = 1.000, p = 0.446, and p = 0.119) with chi-square analysis, although the suspected pulmonary TB patients samples in this study were mostly cough (89.7%), dyspnea (53.8%) and fever (41%). This could be due to cough and shortness of breath are also common symptoms that occur due to respiratory tract disorders, such as lung infections or non-infections³³ whether from the lungs, heart or other causes.³⁴ Fever may be mediated by endogenous pyrogens (cytokines) in response to exogenous pyrogens, especially microorganisms or their products (toxins)³⁵ which can also be caused by organs other than the lungs.³⁶

From the chest x-ray examination in this study showed that each pattern did not correlate with the incidence of pulmonary TB, either infiltrates, consolidations, pleural effusions, cavities and fibrosis (p = 0.446, p = 0.715, p = 0.083, p = 1.000, and p = 1.000) with chi-square correlation analysis. This could be due to the fact that chest x-ray in pulmonary TB can be in several patterns of lession. It differs from size, shape, density, and cavitation. It also can be found in mixed nodular, fibrous, pleural effusion from one person to another.³⁷ In addition, chest x-ray examination in pulmonary TB cannot distinguish whether it is active TB or old TB, so in diagnosing TB, only radiology examination cannot be done.⁵

In addition to Indonesia being a high-burden country contributing to the world TB incidence,¹ TB cases in Indonesia are also increasing every year,⁴ so the Indonesian government is making various efforts to tackle TB cases, one of which is conducting molecular screening tests such as Xpert MTB/RIF Ultra as an initiation of early TB diagnosis in accordance with WHO recommendations.^13,38

The Xpert MTB/RIF Ultra works by detecting MTBC and rifampicin resistance simultaneously by amplifying specific sequences of the rpoB gene using rpo probes 1-4, and the presence of one additional target gene, namely IS1081/IS6110 for detection of MTBC DNA presence in the specimen.¹³ IS6110 is a unique insertion repetitive mobile element specific to the MTBC, which is widely used as a genotypic diagnostic.¹⁵ In addition to IS6110, another gene is IS1081 of the MTBC genome, which is often reported to be very stable and consists of 4 to 6 copies. Several studies have used IS1810 as a NAAT target¹⁴ related to conserved gene regions and multi-copies of specific genes. Xpert MTB/RIF Ultra has a larger DNA amplification reaction space to increase diagnostic accuracy and sensitivity.^12,13 Xpert Ultra has a much higher sensitivity compared to first-generation Xpert in negative acid-fast bacilli (AFB) smear but positive TB culture and HIV-TB patients and shows increased accuracy for the detection of MTBC and rifampicin resistance (RIF).^10,17

Several studies suggest that the sensitivity of Xpert Ultra in diagnosing TB ranges from 92-95% compared to the first generation Xpert (89.40%), and specificity of 96.46% from all lung specimens in pulmonary TB suspects.^39,40 This is different from this study, where the sensitivity of Xpert Ultra was 83.3%, and specificity was 82.5% which is lower than existing literature and some studies. This could be due to there were 28.6% samples (4/14) that did not show growth in MGIT 960 culture system (2 false positive, 2 were NTM identified) that affect these sensitivity and specificity results.

The first false positive patient had received anti-tuberculosis therapy for 4 months from previous TB and then dropped out due to hepatitis B and regularly took antiretrovirals. He also has a history of diabetes mellitus and asthma that is controlled by inhaler. The patient’s clinical symptoms were suggestive of TB, and the lung chest x-ray showed multiple infiltrates and cavities as well as pleural effusion. The second false positive was relapsed pulmonary TB, which received anti-tuberculosis for 6 months and was declared cured in the same year without any acid-fast staining evaluation. The patient presented with shortness of breath and cough, and the patient was also a passive smoker. A chest x-ray showed fibro-consolidation in the right lung and right pleural effusion. From these two false-positive results, could be due to the remnants of non-viable MTB bacteria after receiving therapy. There may be additional potential sources of Mycobacterium tuberculosis DNA in sputum if we consider Tuberculosis as a spectrum of diseases that includes subclinical and even latent stages.⁴¹ Otherwise, the MTB DNA found in sputum can be intact with live or dead MTB, which can be obtained from the lower respiratory system.⁴² Although both results were false positives, TB management was still carried out in accordance with the WHO TB diagnosis algorithm.⁴³

The remaining 2 specimens showed growth on the MGIT 960 culture system but were negative for MPT64 antigen, so they were concluded to be NTM, not MTBC. The first patients is lung adenocarcinoma and multiple-site malignant lymphadenopathy, with the main complaint of a lump in the lymph node accompanied by tightness and weight loss, without respiratory symptoms and TB history. The chest x-ray showed infiltrates in the left parahillary and pleural effusion. The second patient had a hemorrhagic stroke with the main symptoms of dysarthria, she showed respiratory symptoms after a few days of treatment. Chest x-ray radiology showed a right parahilar fibro infiltrate. Of the two patients, IS6110 and IS1081 should only be found in MTBC with positive Xpert Ultra results,¹³ and none of the NTM specimens had these gene targets.⁴⁴ But one of these genes, IS6110, can also be found in NTM,⁴⁵ and another studies state that both genes can be found in NTM,^40,46 so that the Xpert Ultra results were positive and its grew on the MGIT 960 culture system. However, the research is very old, so it is necessary to re-do the research with the same idea or modified.

Of the 28.6% (4/14) patients we described above, 3 of them were trace category and 1 very low detection on Xpert Ultra. When the Xpert Ultra test comes back with a trace result, it’s essential to evaluate many other aspects, according to the WHO. These include the patient’s age, if they have HIV, whether they have ever taken anti-TB medication, and the possibility that the sample’s location is linked to variations in bacterial load. It is necessary to reevaluate clinical symptoms and gather information on previous tuberculosis cases to follow up on the Xpert Ultra trace results. Additional Xpert Ultra tests are not recommended for adults with symptoms of pulmonary Tuberculosis with a positive initial result.¹⁰

In research, specificity and sensitivity may fluctuate with prevalence; thus, considerable disparities in the case between the study and actual practice might lead to spectrum bias in accuracy assessments. However, for a test to be useful, sensitivity combined with specificity should be at least 1.525, or it can be said to be 150%. In this study, we found that the sensitivity and specificity were 83.3% and 82.5%; when summed up, the result is 165.8%, which means that the Xpert Ultra method has good sensitivity and specificity in diagnosing pulmonary TB. Statistically, we conducted the McNemar test in SPSS,²⁶ which is a non-parametric test because the sample in this study included paired nominal data. The differences between the two diagnostic methods, Xpert Ultra and MGIT culture, were found to be insignificant (p = 0.687 by McNemar’s test). This means that the Xpert Ultra method for diagnosing pulmonary TB is the same as the MGIT 960 culture, which is the gold standard for TB diagnosis.

In this study, the positive predictive value (PPV) and the negative predictive value (NPV) were 71.43% and 92%. We assume that the difference in PPV and NPV rates could be due to some samples not detected by culture because the patients had received previous therapy. Positive and negative predictive values are attractive because they are clinically intuitive.⁴⁷ Predictive values can differ from one setting to another for the same diagnostic test, and they are affected by the prevalence of the disease. The lower the prevalence of the disease, the higher its negative predictive value. And the higher the prevalence of the disease, the higher the positive predictive value.⁴⁸

The fundamental metrics of a test’s diagnostic accuracy are sensitivity and specificity. Additional metrics include predicted values and total accuracy. The most crucial factor in the accuracy test is the reference standard that validates the test, and there is the assumption that it is 100% accurate. So, the standard benchmark used in this study is MGIT 960 liquid culture system, the gold standard for TB diagnosis.⁴⁹ So, the closer the accuracy is to 100%, the better the result is. According to this study, the accuracy of Xpert MTB/RIF Ultra for pulmonary TB diagnosis is 84.6%.

According to this discussion, the different results between Xpert Ultra as initial TB diagnosis and culture as the gold standard must still consider clinical judgment as a guide to TB management, the small number of good-quality sputum samples obtained in the study, which can affect the overall accuracy results, is a limitation of this study. Effective coordination between doctors in charge of the patient and the medical technicians in the laboratory administering more than 1 month of anti-tuberculosis therapy can affect the results of Xpert Ultra. Many sputum specimens are sent to the laboratory for Xpert Ultra, but most are not of good quality or are insufficient in volume. Even with the quality of these specimens, positive results can still be produced. In addition, discovering false positive results and NTM findings gives a different view of the sensitivity and specificity values. So, it is necessary to examine the diagnostic accuracy of low-quality sputum, as well as studies that can be focused on those who have never received previous therapy.