PhD Thesis defense by Alberto Gómez Buendía at the VISAVET Centre of the Complutense University of Madrid

February 6^th, 2026

Bovine tuberculosis (bTB), a notifiable zoonotic disease caused by bacteria of the Mycobacterium tuberculosis complex, poses a significant risk to public and animal health. The European Union regulatory framework requires Member States to implement control, surveillance, and eradication programs. Several countries have achieved this objective, but Spain has not yet reached disease eradication despite its efforts. However, many provinces have obtained the officially bTB-free (OTF) status or have very low prevalences, a situation that poses new challenges. These include maintaining the efficiency of the diagnostic protocols implemented, reducing false positives, and avoiding unnecessary culling. This thesis analyses novel epidemiological aspects, focusing on the evaluation of diagnostic test performance and adapting surveillance strategies to optimize control programs at both national and international levels. This approach was framed into three objectives, which subsequently led to the publication of eight scientific papers.

The first objective was to evaluate new methodologies for assessing the performance of bTB diagnostic tests. In this context, Bayesian latent class models (BLCMs) represent a significant advance as they do not rely on imperfect reference tests and allow obtaining less biased estimates of sensitivity and specificity. Thus, they provide a more robust and efficient methodological approach for assessing diagnostic performance. We therefore decided to conduct a review on the use of BLCMs. The review demonstrated that BLCM-based estimates differ significantly from those obtained by reference test-based methods, particularly in antemortem tests: lower sensitivity was observed for skin and serological tests, while a higher sensitivity was found for interferon gamma (IFN-?) release assays. In contrast, results in postmortem tests were more consistent with those obtained by traditional methodologies. This supports the potential of BLCMs to provide more realistic estimates in complex diagnostic contexts.

The second objective pursued the identification of factors related to the infectious agent that could influence the performance of official diagnostic tests. To this end, we evaluated the effect of the M. bovis strain (as determined by the spoligotype) on the probability of lesion detection postmortem. The study, based on a large set of samples from cattle slaughtered in the frame of the eradication program, showed that certain spoligotypes, in particular SB0120 and SB0295, were associated with an increased likelihood of detecting macroscopical lesions while others, such as SB0339, were associated with a lower risk. These findings suggest that the genetic variability of M. bovis strains may influence pathogenicity and thus the performance of certain postmortem tests.

As a contribution to this objective, we also assessed the impact of non-tuberculous mycobacteria (NTM) on the specificity of bTB diagnostic tests. Two studies were conducted: the first identified NTM species from skin test-positive animals from OTF farms in Spain that could be responsible for such non-specific reactions. In this study 373 isolates were analysed and classified into 32 distinct NTM species. Among them, the most common were members of the M. avium complex (MAC), especially M. avium subsp. hominissuis, and (among the non-MAC) M. nonchromogenicum. Subsequently, the ability of six species (M. avium subsp. hominissuis, M. nonchromogenicum, M. avium subsp. avium, M. bourgelatii, M. intermedium, and M. kansasii) selected based on the results of the previous study to induce cross-reactions against different antigens used in official and alternative tests was experimentally evaluated in guinea pig and bovine models. The results demonstrated that MAC species and M. kansasii triggered the most intense responses, particularly when using bTB traditional diagnostic tests based on conventional tuberculins (bovine and avian), although the use of comparative tests (CIT) and alternative antigens (P22 and ESAT-6/CFP-10) led to a reduction in the number of false positives. These results indicate that certain NTM species, especially those in the MAC and M. kansasii, can significantly affect the interpretation of official diagnostic tests, highlighting the need for considering their impact on the design and implementation of the eradication strategies.

The third objective sought to evaluate factors influencing the diagnostic performance of bTB tests, both antemortem (skin test and IFN-?) and postmortem (abattoir surveillance) in low prevalence settings and/or animals from OTF farms. Five studies were conducted with this purpose. The first two focused on factors associated with the presence of skin-test reactors in OTF herds using different methodologies: one used retrospective data from eradication campaigns in Galicia (currently OTF) on a large cohort of animals of more than 3.5 million skin tests in almost 24,000 herds, what allowed the quantification of test performance under field conditions; the other was based on the testing of 3,039 animals from 42 OTF herds in Madrid under more controlled field conditions (all tests performed by the research team), yielding a reference standard to determine the specificity that could be expected in eradication campaigns in high prevalence regions. The retrospective analysis in Galicia showed that the probability of reaction was very low at the individual level (single skin test: 0.052%; comparative skin test: 0.034%), with high herd-level specificity (97.7% and 97.2%, respectively). However, factors such as production type, herd size, season, and region modified the probability of observing skin test reactors. In contrast, the study in OTF herds in Madrid revealed that, even with a low individual probability of reaction (1.24%), the accumulation of reactors generated a considerable proportion of herds with at least one positive animal (70.8%), reducing the estimated specificity at the herd level. Together, these studies indicate that, while skin tests are highly reliable at the individual level, their performance at the herd level in OTF contexts is strongly influenced by epidemiological and operational factors, which require cautious interpretation when prevalence is low.

In low-prevalence settings the antemortem diagnostic pressure is often reduced by increasing the minimum time elapsed between tests conducted in the same herd. While this may lead to improved acceptance of campaigns by the affected stakeholders, it could also lengthen the time period until detection of a bTB outbreak, thereby increasing the risk of disease transmission. Consequently, passive surveillance tools such as postmortem inspection in the abattoir are particularly important in this context, as they compensate for these deficiencies. Given the importance of postmortem surveillance in OTF settings we evaluated the performance of abattoir inspection in Galicia, where between 2014 and 2019 more than 1.7 million cattle were culled in 41 abattoirs. During this period, bTB-like lesions were detected in 223 non-reactors (0.013% of all non-reactors culled) and 301 reactors to antemortem tests (2.3%), with M. bovis infection confirmed in 9.0% and 29.9% of cases, respectively. The probability of detection was influenced by factors such as age (higher in adults than in young animals), production type of the herd (lower in fattening and dairy herds than in beef herds), and province of origin, with a greater variability between abattoirs in non-reactors. Among reactors, the magnitude of the skin test reaction was positively associated with the presence of postmortem lesions. These results highlight the value of abattoir surveillance in complementing reduced antemortem diagnostic pressure in low-prevalence contexts and provide a useful framework for maintaining OTF status.

The official diagnostic test traditionally used in the European Union for the granting and maintenance of such status has been the skin test. However, following the implementation of Regulation (EU) 2016/429, the IFN-? test was also incorporated as an official method for these purposes. Given the limited experience in its application in OTF herds, a study to evaluate its performance under these conditions to assess its usefulness as a tool in eradication programs was conducted. For this, 4,365 samples were collected from 84 farms across five European countries. The samples were tested with two commercial kits (IDvet and Bovigam), comparing different cut-off points and considering animal and herd-level factors. The proportion of reactor animals varied significantly between regions and kits, with higher proportions observed with Bovigam when using cut-offs recommended by the manufacturer. However, ROC analyses showed that many of these discrepancies were attributable to differences in cut-off points rather than to a true discordance between tests. The statistical models applied identified a higher risk of positivity in young animals, in animals from dairy farms, and in those from certain regions, indicating that the specificity of the technique is not homogeneous across populations. By adjusting the cut-off points it was possible to achieve specificities close to 99% in certain contexts, although no universal threshold guaranteed an equivalent performance to skin tests in all situations. These findings suggest that the application of IFN-? as an official test in OTF herds may be feasible in certain scenarios, provided it is accompanied by a prior assessment of baseline reactivity in the target population and by the cut-offs employed.

This collection of eight scientific articles provides a comprehensive overview of the performance of diagnostic tests used in the National Bovine Tuberculosis Program in Spain and identifies the key factors that influence their interpretation and efficacy in a context of low prevalence and OTF.