Tuberculosis (TB) is a kind of bacterial communicable disease caused by the Mycobacterium tuberculosis (MTB) complex (Furin et al., 2019). Spinal TB is a common spine infection disease, usually caused by lymphohematogenous spread from a pulmonary TB and occasionally by genitourinary system TB spread via venous or arterial routes in some cases (Ryu and Patel, 2015; Dunn and Ben Husien, 2018). Spinal TB is prevalent in children and young adults, usually during the initial latent phase of the infection (Ryu and Patel, 2015; Dunn and Ben Husien, 2018). Spinal TB is frequently asymptomatic in its early stage. Gradually, spinal TB can attribute to bone destruction, vertebral collapse, compression of the spinal cord and nerves, and eventually symptoms such as pain, numbness, and weakness (Yong et al., 2021). Therefore, early diagnosis of spinal TB is crucial for preventing the disease and improving therapeutic effects.

To date, spinal TB has been diagnosed by several approaches, including imaging, inflammatory parameters, bone biopsies for microbiological studies or histopathological studies, gene amplification, and immunoassay (Ryu and Patel, 2015; Dunn and Ben Husien, 2018). The various diagnostic methods of spinal TB have their advantages and disadvantages. The imaging presentation of spinal TB is highly consistent with other infections and spinal metastatic adenocarcinoma (Xiang et al., 2021). For inflammatory parameters, the elevation in C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR) has indications but with relatively low specificity (Ryu and Patel, 2015). As the key element for acid-fast bacilli smear (AFBS) (Golden and Vikram, 2005), MTB culture (Dunn and Ben Husien, 2018), molecular polymerase chain reaction GeneXpert MTB/RIF (Chin et al., 2019), and histological exam (Dunn and Ben Husien, 2018), the tissue from bone biopsy has a higher diagnostic yield than pus. Therefore, the biopsy specimen should be as large as possible and include intact pieces of bone, shavings, and necrotic material (Ryu and Patel, 2015). Even so, biopsy-based tests are still less likely to yield positive results. Only 38% of cases were reported by AFBS (Dunn and Ben Husien, 2018), 50% of cases were identified through MTB culture (Lan et al., 2013), and 60% of patients were confirmed through histology from biopsy (Handa et al., 2010). GeneXpert MTB/RIF is a rapid test recommended by WHO with high specificity in the detection of extrapulmonary TB (EPTB) (World Health Organization, 2013) and similar sensitivity for the culture of spinal TB (Patel et al., 2020). Thus, negative microscopy screening for acid-fast bacillus, negative histopathology, negative GeneXpert MTB/RIF result, and failure of M. tuberculosis culture cannot exclude the diagnosis of spinal TB.

Hence, there is an urgent need to find more effective and objective methods for early spinal TB diagnosis. For immunoassay, T-SPOT is a commonly used test for interferon-γ (IFN-γ) release from blood, which is very useful for accurate detection of TB (Luo et al., 2020a). Usually, a high value of T-SPOT means a highly active TB (Janssens et al., 2007). T-SPOT has the potential to improve diagnostic capabilities before surgery. It was also found that MTB-specific antigen (TBAg) to phytohemagglutinin (PHA) ratio (TBAg/PHA ratio) in T-SPOT assay had better predicting capacity in TB (Wang et al., 2016). Therefore, we speculate that the use of combined diagnosis might better distinguish spinal TB from other spinal diseases.

In the present study, we investigated the possible clinical application of T-SPOT assay, GeneXpert MTB/RIF, AFBS, and other methods in the diagnosis of spinal TB. Our results collectively suggested that the TBAg/PHA ratio of T-SPOT assay combined with other methods could be used to establish a differential model for the differentiation of spinal TB from other spinal diseases. Thus, we can establish a new diagnostic model for early spinal TB diagnosis to improve clinical outcomes.

Spinal TB is the most common case of musculoskeletal TB, and specific evidence of culture is difficult to obtain due to the low culture rate of no more than 50% (Lan et al., 2013). The positive culture rate was 41.38% presented in our prospective study. Seriously, an accurate diagnosis of spinal TB takes an average of 1 year and 7 months (Sternbach, 1996). Although imaging (Marais et al., 2018; Jain et al., 2020), laboratory examination (Wang et al., 2019; Jain et al., 2020), histopathology (Thwaites et al., 2009; Jain et al., 2020), and other diagnostic methods have been studied, there is still no applicable diagnostic method to distinguish spinal TB rapidly and early from other diseases. In this study, we proved that the TBAg/PHA ratio showed a better capability to distinguish spinal TB from NTB, especially when used combined with GeneXpert MTB/RIF. To our knowledge, the present study is the first large-scale two-center prospective study embedded in routine clinical practice to assess T-SPOT assay and TBAg/PHA ratio in the evaluation of suspected spinal TB.

It is well-established that the early diagnosis is the basis of anti-spinal TB treatment. Some studies have shown that early control is required to prevent severe persistent kyphosis with attenuation of the spinal cord over the internal gibbus leading to paraplegia (Dunn and Ben Husien, 2018; Jain et al., 2020). As a hematological test, ELISA has developed to be a valuable test for active tuberculosis and latent tuberculosis infection for more than 10 years, whereas the ELISpot (T-SPOT) may be more sensitive (Lalvani and Pareek, 2010). CFP-10 and ESAT 6 are two early secretory proteins of Mycobacterium used in T-SPOT assay, showing their diagnostic ability in various tuberculosis including spinal TB (Kumar et al., 2010). The TBAg/PHA ratio is proposed to be more efficient in distinguishing TB from healthy people (Katakura et al., 2020; Wang et al., 2020). Luo et al. reported that TBAg/PHA ratio alone or combined with iron metabolism index presented a better capacity to distinguish active TB from latent TB (Luo et al., 2020b). Furthermore, Wang et al. also reported that TBAg/PHA ratio improved the diagnostic performance in differentiating EPTB (Wang et al., 2018). In consistent with the previous report, we proved that the diagnostic model of spinal TB based on the TBAg/PHA ratio in the T-SPOT assay had better efficacy, with the AUC value of 0.826 and 0.865 in the training and validation cohort, respectively. In addition, the TBAg/PHA ratio had also rebalanced the sensitivity and specificity of the T-SPOT assay, which might be interfered with the Bacillus Calmette–Guérin vaccine (Yan et al., 2015). Therefore, it suggests that the TBAg/PHA ratio can provide surgeons with preoperative diagnosis recommendations.

GeneXpert MTB/RIF is a rapid and effective PCR test for early spinal TB diagnosis (Patel et al., 2020). AFBS is a traditional and classical method through microscopy examination to diagnose TB in a short time (Vilchèze and Kremer, 2017). GeneXpert MTB/RIF was confirmed to be advantageous for TB diagnosis, particularly extra-pulmonary TB with negative AFBS (Mechal et al., 2019). Gerardo et al. reported that GeneXpert outperformed AFBS in all types of tuberculosis especially in cerebrospinal fluid (Alvarez-Uria et al., 2012). In this study, we found that GeneXpert MTB/RIF, with an AUC of 0.750, had a much better diagnostic efficiency than AFBS with an AUC of 0.575, in the training cohort. The infected bone tissues are superior to pus in the diagnosis of spinal TB, which is also confirmed in our TB culture results, consistent with Swapna report (Kanade et al., 2019). However, the positive rate of GeneXpert was <60% in our results because the diagnosis of spinal TB depends on the precise biopsy by the spine surgeon. Therefore, the diagnosis of spinal TB with GeneXpert MTB/RIF alone cannot meet the diagnostic requirements.

Since the T-SPOT is an assay based on immunity, we assumed that GeneXpert MTB/RIF could be a complementary method combination in the diagnosis. In our results, the TBAg/PHA ratio was proved to be superior to the T-SPOT results in diagnosing spinal TB. The diagnosis model based on the TBAg/PHA ratio combined with GeneXpert MTB/RIF significantly increased the AUC both in training and validation cohorts. The combination also had higher diagnostic PPV, NPV, PLR, NLR, and accuracy. Thus, the addition of GeneXpert MTB/RIF to the TBAg/PHA-based diagnostic model could improve the capacity to distinguish spinal TB from NTB. In other words, patients suspected of spinal TB need to undergo two tests, namely, the T-SPOT test and the GeneXpert MTB/RIF test. After the completion of the T-SPOT test, the predicted value of the TBAg/PHA ratio could be used for early warning of spinal TB. Then, a careful precise biopsy should be performed with skilled spine surgeons. If GeneXpert MTB/RIF results are negative, a second biopsy might be required when the TBAg/PHA ratio is much higher than the cutoff value. First, by calculating the prediction level of the combination of the TBAg/PHA ratio and GeneXpert MTB/RIF, it can guide early treatment when waiting for the gold standard diagnosis of spinal TB. For example, if a patient’s predictive value after the calculation is four times greater than the cutoff value (specificity >95%), combined with highly suspicious imaging, the patient is highly likely to have spinal TB. Then, preventive antituberculosis therapy is recommended before the gold standard diagnosis to alleviate symptoms, speed recovery, and prevent a postoperative recurrence. Second, the gold standard of EPTB has the problem of high missed diagnosis (Norbis et al., 2014), and preventive anti-tuberculosis treatment is sometimes required. Our model can provide preventive treatment suggestions for patients with the negative gold standard diagnosis but highly suspected spinal TB. As the diagnosis of spinal TB is complicated (Wang et al., 2019), our model also suffers from misdiagnosis and missed diagnosis. Thus, patients need to complete the gold standard test to provide conclusive evidence for subsequent standard treatment. In addition, our model has other shortcomings, which may be interfered with by latent TB. The diagnosis of latent TB is complicated. CT scan, clinical examination, medical history inquiry, and physical examination can be utilized to avoid the inclusion of latent TB. As a result, latent tuberculosis after exclusion has little effect on our research. How latent TB influences the outcome is an issue that needs further exploration in the future. GeneXpert machines are expensive and not easily accessible in some developing countries. Although the TBAg/PHA ratio combined with GeneXpert MTB/RIF had better performance than the TBAg/PHA ratio with AFBS in distinguishing spinal TB from NTB, the performance of the latter is better than that of TBAg/PHA alone. Considering the relatively simple technology and low price, the combination of TBAg/PHA ratio with AFBS could be also applied in underdeveloped areas or population.

It is widely known that spinal TB is most prevalent in developing countries and socioeconomically deprived communities. Other precise diagnostic methods may also be conducive to the individualized diagnosis of spinal TB. The real-time high-resolution imaging of TB colonies is a new technology that shortens the time to growth detection in culture (Ghodbane et al., 2015). TBDx system (Signature Mapping Medical Sciences, Herndon, USA) automatically loads slides onto a microscope and uses computerized algorithms to classify auramine-stained smears with a significantly higher sensitivity than manual microscopy (Lewis et al., 2012). Over 300 individual novel TB antigens including Rv2031c, Rv2029c, antigens of the Ag85 complex, and Rv0475 were reported in previous tests (Meier et al., 2018), providing other feasible strategies in the future. QuantiFERON-TB Gold (QFT) is also an IFN-γ release assay that might replace T-SPOT in the diagnosis of spinal TB (Acharya et al., 2020). Other nucleic acid probes and gene amplification techniques including loop−mediated isothermal amplification, line probe assay, or other gene-specific PCR (Acharya et al., 2020) could be used to investigate the diagnosis capacities of combined TBAg/PHA ratio. Studies of T-cell immune response (Jasenosky et al., 2015), subgroup immune cell changes (Cardona and Cardona, 2019), and genome sequencing (Huang et al., 2020) will be another way to distinguish spinal TB from NTB.

Notably, the present study had some issues to be addressed. First, to keep the independence of the training and validation cohorts, the sample size of the training cohort is relatively small. Second, the recruitment of patients was based on suspected spinal TB on radiological consideration. There existed missed diagnoses in radiologists’ reports and a bias between different radiologists. Third, since spinal TB is paucibacillary lesions, the fluoroscopy-guided percutaneous bone biopsy was limited by the disease itself. Although GeneXpert MTB/RIF relies on tissues from biopsy or operation, tissue dependence also occurs in MTB culture, AFBS, and histopathology. Fourth, although we try to avoid the inclusion of latent TB, it might interfere with the experimental results.

In conclusion, this study successfully provided generable evidence for the TBAg/PHA ratio of T-SPOT detection in distinguishing spinal TB from NTB in routine clinical application. The diagnostic model based on the combination of TBAg/PHA ratio with GeneXpert MTB/RIF possessed more effective results for spinal TB diagnosis.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology. The patients/participants provided their written informed consent to participate in this study.

YQ, YL, and FL co-designed the study. YQ wrote the first draft of the manuscript. XL, ZF, and YL wrote sections of the manuscript. YQ, ZL, ZF, YL, and FL performed research. YQ, ZL, and YL performed the statistical analysis. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.