Epidemiology Aspects of Brucella-DNA Detection Using Real-time PCR from Bovine Whole Blood Samples in Colombia

A cross-sectional study was conducted in Colombia to recover Brucella spp. DNA from bovine whole-blood samples through probe-based real-time PCR (Probe-qPCR) and by an SNP-based assay, differentiating vaccine strains from eld strains. The associated factors for the presence of Brucella-DNA were reported and evaluated using logistical regression models. A total of 656 random cows from 40 herds were selected. Template DNA was obtained based on a modied salting-out protocol. The Probe-qPCR assay using bcsp31 gene amplication showed an eciency of 92.35%, with a slope of -3.52 reached in the standard curve. The qPCR assay detected 9.5% (n = 62/656; 95% CI: 7.3,12.0) of the animals with Brucella-DNA presence, and 62.5% (n = 25/40; 95% CI: 45.8,77.3) of the herds with Brucella-DNA presence. Using the SNP-based assay, all positive samples were identied as eld Brucella strains. In the nal regression model at the animal-level, ve variables were associated with Brucella-DNA presence: the use of bulls for mating, recorded history of reproductive problems, pregnant cows, parlor milking, and cows belonging to farms ≤ 200 m from the main road. At the herd-level, two variables were associated with Brucella-DNA presence: recorded history of reproductive problems and bulls' use for mating. Given the uctuant brucellosis prevalence in endemic areas, updated epidemiological studies are necessary to evaluate the disease dynamic, and if established prevention and control measures have been effective or need to be adjusted. The increase in the prevalence of brucellosis in animal reservoirs creates an important risk of transmission in humans.


Introduction
Brucella spp. are the causative agents of brucellosis, a complex and worldwide zoonotic disease that has been neglected for a long time. According to the Plan of Action for eliminating Neglected Infectious Diseases (NIDs) 2016-2022, there continues to be a need to evaluate and document the regional epidemiological situation concerning brucellosis for the Americas. Although presenting endemic brucellosis, Latin America reports insu cient data on the prevalence and has not yet reached satisfactory levels of disease control (World Health Organization and Pan American Health Organization, 2016).
Depending on the prevalence found, the strategies and control measures could apply different management actions (Moreno, 2020). Furthermore, proper detection can help mitigate the economic impact generated by the culling of infected animals (Trotta et al., 2020), the costs to recover a brucellosisfree status, the socio-economic impact on rural development, and more importantly, the public health impact.
Bacterial culture is considered the "gold standard" to con rm brucellosis' diagnosis. Nevertheless, most Brucella species are fastidious and relatively slow-growing organisms, affecting in some cases, the detection sensitivity (World Organization for Animal Health, 2018). Instead, epidemiological studies targeting brucellosis in livestock are based almost exclusively on serology tests as the main detection tools due to their rapid results and low cost. However, serological tests present low sensitivity in early or latent stages of infection and lack speci city in regions where brucellosis is highly endemic, other bacteria share the structure of O-LPS, or there is massive cattle vaccination (Godfroid et al., 2002;Corbel, 2006). In countries with massive vaccination, the reliable differentiation between vaccine strains and eld strains of Brucella sp. is crucial for epidemiological studies of bovine brucellosis. (Gopaul et al., 2010). Therefore, better tests and testing strategies, adapted to the epidemiological situation, need to be developed and validated (Godfroid et al., 2002).  (Hull et al., 2018), combining speed, sensitivity, high speci city (with use of probe), laboratory safety, and low risk of cross-contamination or amplicon contamination in the bloodstream (Corbel, 2006). The qPCR technique's application may support the drawbacks of the indirect methods and the low sensitivity of the microbial isolation technique. This cross-sectional study evidences the associated bovine brucellosis factors from Brucella DNA from bovine whole blood, differentiating vaccine strains from eld strains by an SNP-based assay.

Sampling and target population
The sampling procedure applied the formula n = 1.962 p (1 − p)/d2 to calculate the sample size with a con dence level of 95%, desired absolute precision (d2) of 5%, and expected prevalence (p) of 10% (Thrus eld, 2007). A total of 656 whole-blood samples were randomly collected from cows in 40 dairy herds of four municipalities in the north of Antioquia. Proportionate sampling was used from each city to ensure the number of dairy cattle has not been underrepresented.
The target population was composed of animals in lactating or in the dry period, ≥ 2-year-old, vaccinated at the age of 3 to 8 months, mainly with 10-34 x 10 9 colony forming units (CFU) of rough attenuated strain RB51 or alternatively with 20 x 10 9 CFU of smooth live vaccine strain S19. Mandatory vaccination was veri ed by o cial vaccination records (RUV) (Instituto Colombiano Agropecuario, 2017). The animal's data was taken from the Foot and Mouth Disease and Brucellosis vaccination campaign (Instituto Colombiano Agropecuario, 2018a). The herds' selection was made considering the producers' availability to participate voluntarily in the research and signed the informed consent agreement. Each farmer gave a full animal list to select the participants. A random number table without replacement was used in the cow's election. From herds with ≤ 50 cows, 15 dairy cows were chosen for the current research. From herds with > 50 cows, 22 dairy cows were selected. The randomly chosen animals were transferred to a designated handling area within each farm for collecting the samples. The animal skin area was cleaned with chlorhexidine and six milliliters of caudal venous blood were taken from each animal using blood collection tubes with ethylenediaminetetraacetic acid, an anticoagulant. The blood samples were refrigerated and properly transported to the lab.

Epidemiological data collection
Dairy farmers were asked to answer a questionnaire. The questionnaire was based on a literature review and included closed questions regarding cattle management, zootechnical parameters, and sanitary practices as independent variables (Tables 1 and 2). The questionnaire was pre-tested and validated by farm managers that did not belong to the study.

DNA extraction from whole blood samples and quality analysis
Template DNA was obtained based on the non-organic salting out protocol designed for the recovery of ). Blind sample processing was used, assigning a code to each sample, to avoid potential observer bias.

qPCR standard curve
The concentration of DNA extracted from the B. abortus strain present in the S19 vaccine (Vecol S.A, Colombia) was used to calculate the number of genome copies, assuming a genome size of 3.28 x 106 bps. The standard curve was built using six serial decimal dilutions of the puri ed template DNA (from

Statistical analysis
The epidemiological data was analyzed using the statistical program IBM SPSS (Statistical Package for the Social Sciences) v. 25.0 for Windows (Armonk, Nueva York: IBM Corp.). Pearson correlation coe cient was used for quantitative variables with the normality assumption. Spearman's rank-order correlation was applied when quantitative variables did not meet the normality assumption. A chi-square test or Fisher test was used to evaluate the association between Brucella-DNA presence by qPCR assay with all the explanatory variables. Statistically signi cant variables and variables with p-value < 0.25 were offered to the logistic regression models from individual animal and the herd level. Before creating the nal model, the multicollinearity was measured by determining the variance in ation factors (VIF) and tolerance. Collinearity was considered when the VIF value exceeds 6.0, and the tolerance was less than 0.2. According to the literature review, when two variables were found with high collinearity, only the variable with more biological importance and plausibility was included in the model. The model was built by the "intro" method. Finally, the coe cients were estimated, and the appropriateness adequacy, validity, and usefulness of the model were identi ed.

Results
Descriptive analysis exhibited that the mean age of evaluated cows was 59.2 ± 24.6 months old. The mean gestational age was 81.3 ± 84.5 days, and the offspring's mean was 3.0 ± 1.  (Fig. 2). The SNP-based assay con rmed that all positive samples were from Brucella spp. eld strains.
The quantitative variables did not meet the normality assumption and were not statistically signi cant by Spearman's rank-order correlation. Therefore, they were transformed into ordinal variables in the model (Tables 1 and 2

Discussion
Brucellosis is a "tool-de cient" disease that has not yet reached satisfactory levels of disease control (World Health Organization, 2007). The detection of antibodies provides only a provisional diagnosis, but there may also be an exceptionally long incubation period in some infected animals, and individuals may remain serologically negative for a considerable period following infection (Corbel, 2006  it is necessary to update the epidemiological situation analysis to implement effective control and prevention strategies. This study is the rst report of molecular prevalence in the region, showing a Brucella-DNA prevalence in concordance with the declared outbreak and higher than the seroprevalence records in the region. Bovine whole blood as the sample for PCR detection of Brucella spp. has been little considered because Brucella may replicate at low frequency, leading to low and short bacteremia periods (Vrioni et al., 2008). However, under experimental conditions, it has been shown that some Brucella species could invade red blood cells and settle in the cytoplasm without replicating, establishing persistent bacteremia with different bacterial loads (Vitry et al., 2014;Gwida et al., 2016). In humans, a study of three patients infected in the recent Brucellosis outbreak in China con rmed that Brucella persists in peripheral blood (Baoshan et al., 2020). In this study, a high recovery of Brucella-DNA from whole bovine blood was possible. A high Brucella-DNA presence from bovine blood by qPCR, was also obtained in a bovine brucellosis outbreak study in Egypt (Gwida et al., 2016).
The Brucella-DNA presence could indicate the viable bacteria's existence in the cattle blood, but there is also a possibility that the presence of Brucella-DNA could be associated to particle shedding by dead bacteria or vaccine-DNA (Vrioni et al., 2008). The main used vaccine in the studied region is the RB51, however, the presence of the RB51 vaccine strain in the blood should not generate alerts. According to The Animal and Plant Health Inspection Service (APHIS), the RB51 must be rapidly cleared from peripheral blood within three days post-vaccination (National Council of State Legislatures, 2018). The In dairy herds, mainly in large herd sizes, there is a practice of raising calves and heifers in a different property from milk production. Some studies reported that animals' mobilization is a risk factor for introducing brucellosis in herds free of the disease (Cárdenas et al., 2019;Corbel 2006). Since these animals are exposed to different management practices and environmental conditions, this exposure can favor brucellosis infections (Corbel, 2006;Wareth et al., 2015). In this study, no signi cant association of the presence of Brucella-DNA with the raising of calves in other farms was found, but a signi cant association was found with the transfer of animals. Raising calves and heifers at a different property from the dairy cows can be appropriate if managed correctly. Separating and keeping the calves and heifers isolated from other herds, then testing them appropriately prior to their return will prevent Brucella introduction into dairy herds (Instituto Colombiano Agropecuario, 2017).
In the logistic regression model, the use of bulls for mating explained the Brucella-DNA presence at the animal and the herd level in the present investigation. Other epidemiological studies also showed a signi cant association of the bulls with Brucella seropositivity. They considered that sharing a stud bull with neighbors could likely be a potential source of Brucella infection (Muma et al., 2007;Ahasan et al., 2017). Brucella spp. can be found in the testicles and the male genital glands and transmitted by infected semen (Corbel, 2006). However, differences in vaginal acidity, mucosal immunity, and cervical mucus between estruses and anoestrus may affect bacterial survival and kill the bacteria (Uhrig et al., 2013). Therefore, it is more likely that females' infection may occur through environmental contamination with infected seminal discharges (Thomsen, 1943). In Yellowstone National Park, USA, a study showed seroconversion in four of eight bison after intravaginal inoculation suggests that B. abortus transmission by the venereal route in bison can occur. Although routes of transmission are considered like those in cattle (Uhrig et al., 2013), older experiments did not nd an essential role in bulls' use for mating in Brucella transmission in cows during copulation (Thomsen, 1943). Updated studies on venereal transmission in cattle are needed. Meanwhile, all bulls from known infected herds should be viewed with suspicion regarding their Brucella spp. infection status (Plant et al., 1976).
Management practices, mainly associated with reproduction, are far more critical in determining the risk of Brucella spp. infection in bovine (Corbel, 2006). Variables such as the history of reproductive problems, pregnant cows, days of gestation, and disposal of birth waste were associated with the Brucella-DNA presence in the animals evaluated. Pregnant cows have been associated with the probability of Brucellaantibody presence in a cross-sectional study in Ethiopia and a study conducted for isolation and subsequent detection by PCR of Brucella spp. in milk samples from seronegative cows in Bangladesh, even showed prevalence almost double in pregnant cows than in nonpregnant cows (Islam et al., 2018).
Apart from the last two, these factors were associated with Brucella-DNA presence in the logistic regression models.

Conclusion
This study found the recovery of Brucella-DNA from whole bovine blood was possible and considered it a clinical sample for Brucella-DNA detection and highlighted the complementary role of PCR techniques in detection, provided a highly sensitive molecular test is used.
Permanent surveillance of brucellosis in livestock is imperative. Several epidemiological aspects obtained from the molecular prevalence evidenced in this study are in concordance with other seroprevalence studies. However, this research retakes unconsidered associations, like the bull's role in the brucellosis transmission, and suggests the need for spatial studies and transmission models in environments. The current research also highlights that management practices, mainly associated with reproduction, favors the Brucella infection and its propagation inside and outside herds.
Given the uctuant brucellosis prevalence in endemic areas, updated epidemiological studies are necessary to evaluate the disease dynamic, and if established prevention and control measures have been effective or need to be adjusted. Figure 1 The standard curve was plotted using six serial decimal dilutions of Brucella abortus S19 vaccine DNA (from105 and 0.1 copies). The quanti cation cycle value was plotted as a function of the initial number of copies.