Exploring the presence of bovine leukemia virus among breast cancer tumors in a rural state

The bovine leukemia virus (BLV) is a deltaretrovirus that causes malignant lymphoma and lymphosarcomas in cattle globally and has high prevalence among large scale U.S. dairy herds. Associations between presence of BLV DNA in human mammary tissue and human breast cancer incidence have been reported. We sought to estimate the prevalence of BLV DNA in breast cancer tissue samples in a rural state with an active dairy industry. We purified genomic DNA from 56 fresh-frozen breast cancer tissue samples (51 tumor samples, 5 samples representing adjacent normal breast tissue) banked between 2016 and 2019. Using nested PCR assays, multiple BLV tax sequence primers and primers for the long terminal repeat (LTR) were used to detect BLV DNA in tissue samples and known positive control samples, including the permanently infected fetal lamb kidney cell line (FLK-BLV) and blood from BLV positive cattle. The median age of patients from which samples were obtained at the time of treatment was 60 (40–93) and all were female. Ninety percent of patients had invasive ductal carcinoma. The majority were poorly differentiated (60%). On PCR assay, none of the tumor samples tested positive for BLV DNA, despite having consistent signals in positive controls. We did not find BLV DNA in fresh-frozen breast cancer tumors from patients presenting to a hospital in Vermont. Our findings suggest a low prevalence of BLV in our patient population and a need to reevaluate the association between BLV and human breast cancer.

BLV is a deltaretrovirus, first isolated in 1969, and subsequently demonstrated to be the causal agent of enzootic bovine leucosis (EBL) characterized by B cell lymphocytosis, lymphoma, and malignant lymphosarcomas [54,60,61].BLV encodes a viral trans-activating protein, tax, which is required for the experimental induction of malignancy and alteration of host cell gene expression and proliferation [62,63].In the U.S. approximately 38% of beef herds, 84% of dairy herds, and 100% of large-scale dairy operation herds are infected with BLV [64,65].
Several studies report positive associations between BLV DNA in breast tissue and incident breast cancer.In a case-control study analyzing breast tissue from patients in multiple U.S. states, BLV DNA was found in the mammary epithelium of cancer patients more often than in cancer-free controls (multivariable odds ratio: 3.07 (95% CI 1.66, 5.69) [55].In an Australian case-control study, BLV DNA was found in 80% of women with breast cancer versus 41% of women with no history of breast cancer, resulting in an age-adjusted odds ratio of 4.72 (1.71, 13) [56].A recent study of breast tissue collected at MD Anderson (Houston, Texas) likewise demonstrated an increased association of BLV tax DNA in the breast tissue of women diagnosed with breast cancer and reported an age/ ethnicity-adjusted odds ratio of 5.9 (2.8, 12) [57].A recent systematic review and meta-analysis of 9 studies from the U.S., Australia, Iran, Colombia, and Brazil found that identifying BLV infection conferred an increased odds of breast cancer (OR = 2.57, 95% CI 1.45, 4.56) [59].Despite the associations reported in these studies, there is also data to the contrary.Gillet et al. performed whole genome sequencing of 51 cancer tissue samples and adjacent mammary tissue obtained from the NCBI Database of Genotype and Phenotype (dbGaP) and did not find any pairing with BLV variants [66].Adekanmbi et al. assessed presence of BLV in 238 blood-derived samples from breast cancer patients.Using both Fluorescence Resonance Energy Transfer PCR and whole genome sequencing in a subset of patients, no BLV proviral DNA was detected [67].Similarly, Zhang et al. was unable to detect BLV in breast cancer tissue among 91 samples from Chinese women from provinces with reported high prevalence of bovine BLV infection [68].Thus, the potential role of BLV in human breast cancer remains unclear and controversial.
Between 2009 and 2013, Vermont females had a slightly higher incidence of breast cancer (128.3 cases per 100,000 population) compared with U.S. females overall (122.9 cases per 100,000 population) [69].However, the prevalence of BLV in Vermont female breast cancer has not been investigated despite the state's large dairy cattle industry.The objective of this study was to estimate the prevalence of BLV DNA within breast cancer tumors from patients residing in Vermont through analysis of fresh-frozen tumor tissue from our local tumor biobank.

Methods
This study was approved by the University of Vermont Cancer Center Protocol Review and Monitoring Committee and the University of Vermont Institutional Review Board (Study ID: 00000564).

Sample acquisition and data collection
Tissue samples were obtained from the University of Vermont Cancer Center Tissue Bank, under IRB approved protocols for tissue procurement and biobanking.Samples are linked to clinical and pathology data from electronic medical records.We identified patients with a diagnosis of invasive breast cancer who underwent surgical excision between 2015 and 2019 and had adequate tissue banked for analysis.We excluded patients who received neoadjuvant chemotherapy.We collected fresh-frozen breast tumor samples from 30 patients, selected at random from the set of 309 eligible cases.All patient samples included tumor tissue, and 16 also included adequate amounts of adjacent fresh-frozen normal tissue from the operated breast.Basic demographic, clinical, and tumor data were collected for all 30 breast cancer patients including age, sex, histologic subtype, grade, receptor status, and stage.To assure that the samples came a representative cohort of patients, we also collected the same demographic, clinical, and pathologic data on an additional 30 patients, selected at random from the set of eligible cases, but whose tissue was not analyzed.

Sample processing and evaluation
Tissue samples (approximately 25 mg per sample) were thawed on ice and then subjected to genomic DNA purification using the MagMax DNA Multi-Sample Kit (Applied Biosystems, ThermoFisher Scientific, Waltham, MA) according to manufacturer's recommendations.Genomic DNA from seropositive and seronegative bovine whole blood samples (300 µL) were purified in the same manner.Whole blood samples were obtained from cattle from the University of Vermont teaching and research herd in accordance with Institutional Animal Care and Use Committeeapproved procedures (protocol 08-099).At the time of sampling this herd had a known history of enzootic bovine leukosis (28% BLV positive by enzyme-linked immunosorbent assay (ELISA) testing annually).Individual serum samples of adult lactating cows were submitted to the Cornell University College of Veterinary Medicine Animal Health Diagnostic Center (Ithaca, NY) for ELISA testing.Whole blood samples (1 mL aliquots) were collected in parallel with serum collection and stored at either 4 °C or −20 °C until DNA extraction.Additional positive control genomic DNA was prepared from the persistently BLV-infected fetal lamb kidney cell line, FLK-BLV [70].Extracted DNA concentrations were measured using a NanoDrop 2000 spectrophotometer (ThermoFisher Scientific, Waltham, MA).Quality of extracted human DNA was confirmed by amplification of human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) genomic sequences [54].
Nested PCR (nPCR), as described by Buehring et al., was used to detect BLV sequences in extracted DNA samples [54].Briefly, extracted genomic DNA template (100 ng) was combined with 0.2 µM of outer amplicon forward and reverse primers (Table 1 and Fig. 1) in a reaction mixture containing 0.4 mM dNTPs, 4 mM MgCl 2 , and 0.05 U/µL of Taq Polymerase (supplied as DreamTaq Hot Start Green Mastermix, ThermoFisher Scientific, Waltham, MA) in a final volume of 50 µL.Cycling parameters were consistent with those described by Buehring et al., with the addition of an initial hot start step (3 min at 95 °C) prior to cycling to activate the DreamTaq Hot Start enzyme [54].Upon completion of cycling of the outer amplification reaction, 2 µL of outer nPCR reactions was removed and added to 48µL of  PCR reaction mix containing 0.2 µM of inner amplicon forward and reverse primers.Amplification of target sequences was monitored by 1% Agarose TBE electrophoresis, loading 20 µL of inner amplification reactions and running at 80 V for 60 min.Gels were imaged on a BioRad ChemiDoc XRS digital imaging system and analyzed using BioRad ImageLab software (BioRad, Hercules, CA).
To establish tax nPCR sensitivity, FLK-BLV gDNA was serially diluted (1 ng per reaction, diluted fourfold) and subjected to nPCR (tax Amplicon A), as described above.Based on these sensitivity evaluations, 50 pg of FLK-BLV genomic DNA was used as a template for positive controls.(Supplemental Fig. 1).

Statistical analysis
Initial sample size requirements were estimated based on reported proportions of BLV positive patients in Beuhring et al. [55], which estimated 30% BLV positive in the control group and 60% positive in the cancer group.Our sample size of 30 afforded 99% power to detect a prevalence proportion of 30% against a null hypothesis of 1%.
After initial samples from 20 patients yielded no positive results, we calculated the probability of receiving all negative results with conservative assumed true prevalence of 10%, and determined that testing 30 samples would yield an approximately 4% probability of observing all negative results.Under an assumed true prevalence of 30%, this probability was 2.3 × 10 -5 .

Results
The median patient age was 60 (range: .A second randomly-chosen control cohort of 30 patients with samples available from the biobank who fit the inclusion/ exclusion criteria but who did not have their samples were also assessed for their clinical characteristics and demonstrated that experimental cohort was relatively similar to the patients whose samples were assessed for BLV DNA expression.The majority (90%) of tissue diagnoses were invasive ductal adenocarcinoma, with the remaining classified as invasive lobular carcinoma.Three (10%) were ductal carcinoma with lobular features with one sample diagnosed as both invasive lobular and ductal adenocarcinomas (Table 2).This comports with estimates from the American Cancer Society that approximately 80% of invasive breast cancers present as invasive ductal carcinoma [71].
In our cohort, 36.7% of patients had regional disease, which was slightly higher than our control cohort and reported rate in Vermont of 20% [69] and may reflect the more aggressive histology seen in our sample compared with our control cohort.Histologically, 59.3% of tumors were classified as poorly differentiated, 33.3% as moderately differentiated, and 7.4% as well differentiated.Estrogen, progesterone, and HER2 receptor positivity was found in 90.0, 86.6 and 66.7% of tumor samples, respectively (Table 2).This compares with the reported rates in Vermont in which 78% of breast cancers express hormone receptors and 13% overexpress HER2 with 9% of tumors classified as triple negative expression.Therefore, our sample tended towards more hormone receptor positivity and HER2 overexpression compared with the reported rates [69].
Nested PCR detected no BLV DNA sequences in all samples from 30 patients-whether from tumor or adjacent tissue-despite achieving positive signals in positive control specimens (FLK) and in blood samples from BLV-ELISA positive cows (Fig. 2).Known negative controls, including blood samples from BLV-ELISA negative cows, tested negative as expected.Based on sensitivity evaluations, a positive result was achieved requiring less than 15 pg of template for detection (Supplemental Fig. 1).This limit of detection, based on the observation that the sheep genome is ~ 2.61 Gbp (2.86 pg per cell) [72] and each FLK-BLV cell contains 4 proviral copies [70], corresponds to between 2 and 10 copies of provirus target, equivalent to approximately a single infected cell or a small number of cells containing provirus in a tumor mass.
In assessing BLV strain phylogeny from Polat et al., we identified prominent strains isolated from the U.S. and performed sequence alignments of these isolates using Clustal Omega (European Molecular Biology Laboratory's European Bioinformatics Institute).[73,74] To account for polymorphisms in critical priming regions, we applied different primer combinations in nested PCR as indicated in Table 1 and Fig. 1.This included primers to amplify both the tax sequence as well as the viral LTR sequence.Similarly, no positive cases were detected.
Our initial study planning, based on values in published studies, assumed prevalence of BLV in breast tumors of 30%.Even at an assumed 10% true prevalence, we calculated that with our sample size of 30, there would only be a 4.2% probability of having no samples positive for BLV by chance alone.
Given the lack of detection of BLV in our samples, we could not carry out analyses associating presence of BLV DNA with any of the clinical or pathologic features of the breast tumors.

Discussion
Prior literature has suggested an association between BLV and breast cancer risk.The published record includes a systematic review and meta-analysis, [59] and multiple case-control studies [54,55,[55][56][57][58].However, similar studies have concluded that there is no association of BLV with human breast cancers [66][67][68].We therefore sought to determine the presence of BLV in breast cancer tissue in Vermont, a highly agriculture state with an active dairy industry and an established presence of BLV in cows in the region.
Basing our methods on previously documented approaches [54], we initially tested tumor samples for Fig. 2 Nested PCR results for breast tissue samples tumor (T) and normal tissue (N) by primer utilized.A Primer pairs specific for endogenous human GAPDH were used to determine genomic DNA quality obtained from tumor and normal breast tissue samples (i.e., ability to amplify DNA in sample) and B subsequent primer pair combinations as defined in Table 1 were used to generate specific BLV tax and LTR amplicons.All patient samples were analyzed twice, only single results shown.FLK denotes the FLK-BLV gDNA positive control, NTC denotes the no-template negative control.C Nested PCR results performed on DNA extracted from bovine whole blood samples demonstrated amplification of the tax amplicon in samples known to be seropositive for BLV infection and no amplification in the samples known to be seronegative amplification of the BLV tax gene.This gene was chosen because trans-activating region of the X-gene (tax) is thought to play a mechanistic role in oncogenesis [75].Furthermore, the genomic region occupied by tax is highly conserved with little variation in sequence [76] and tends to remain detectable even in advanced phases of the disease when other regions of the viral genome are deleted [77].Therefore, the viral tax DNA sequence that has been inserted into the host genome provides a stable target for assessing the presence of mammary cell infection [57].
Since our tissue samples showed no amplification of BLV genetic signatures, we sought to address any technical reasons for this lack of signal.To account for the possibility that these results could be due to difference in the particular tax gene sequence preventing identification of the tax gene with the initial primers used, we assessed BLV strain phylogeny from Polat et al., to identify prominent strains isolated within the U.S. and performed sequence alignments of these isolates using Clustal Omega (European Molecular Biology Laboratory's European Bioinformatics Institute) [73,74].We applied different primer combinations as outlined in Table 1 and Fig. 2 to account for polymorphisms in critical priming regions that could have accounted for our initial negative results.We furthermore tested for BLV LTR sequence, which plays an important role in controlling viral gene expression [78] and has been shown, along with the tax region, to be highly conserved [79][80][81][82].
Consistently negative results using this variety of primers for tax and LTR along with confirmatory testing of our positive and negative controls and the finding of expected results in known seronegative and seropositive bovine blood suggest that our assay was both sensitive and specific for BLV.In particular, we were able to accurately identify presence/ absence of BLV in locally-sourced bovine samples, which would most likely reflect the genetic variation of BLV that could be identified in the corresponding human population in Vermont.One potential limitation is that no positive control was based on a fresh-frozen tissue source.However, DNA concentrations were confirmed for extractions from different sources (cell lines, bovine blood, and fresh frozen tissue) and DNA was purified in a manner that we would not expect to be sensitive to type of source material.The quality of extracted tissue DNA was successfully confirmed by amplification of the gylceraldehyded-3-phosphate dehydrogenase (GAPDH) gene sequence.Furthermore, use of frozen tissue for viral DNA quantification by PCR has been reported and validated in the literature.Davidson F et al. assessed how HIV-1 DNA extracted from frozen vs. formalin fixed paraffin embedded (FFPE) brain tissue was detected by PCR and correlated with viral load.They found a lower rate of detection in FFPE whereas frozen tissue was found to have a better correlation with viral load [83].A higher viral DNA yield by PCR from frozen tissue as compared with FFPE has been supported [84], and other studies have likewise described detecting and quantification of viral DNA from fresh frozen tumor samples [85].
Additional confirmatory testing could be considered using other modalities aside from PCR, such as immunohistochemistry (IHC) or in situ hybridization (ISH), but neither of these methods approaches the detection limits of nPCR.IHC also requires viral antigens and would be predicated on active infection, while nPCR detects either active or latent BLV [86,87].While a novel technique using ISH to identify AS1 (a non-coding RNA derived from the BLV genome) has been described on formalin-fixed paraffin-embedded (FFPE) tissue samples from cattle infected with BLV, this technique has not yet been validated in human mammary tissue [87].
Another potential limitation of this study is its low sample size and the corresponding possibility that we achieved all negative results by chance.However, assuming no measurement error, there is a very low probability that all the breast cancer tissue samples would be truly negative under a variety of plausible prevalence assumptions.For example, assuming 30% BLV DNA prevalence in breast tumor tissue, the binomial probability of obtaining 30 negative specimens is ~ 0.0023%.Under an assumed prevalence of 10%-which is much lower than prevalence estimates from earlier studies-the binomial probability of obtaining 30 negative specimens is 4.2%.This could indicate a lower true prevalence in Vermont than that reported from other locales, again assuming no measurement error.Alternatively, it could suggest that early studies have overestimated the prevalence in breast tissue and therefore that our initial assumptions should have been based on a lower prevalence of BLV DNA in breast tissue.In order to minimize that there was a selection bias, samples were randomly chosen and clinical characteristics compared to another set of 30 randomly chosen patients with samples available that were not texted.Cohorts were similar in most clinical characteristics, thought with a slightly higher prevalence of poorly differentiated specimens and those with regional disease.It is unclear that this difference would have any association with or mechanism for BLV expression or lack thereof.
While we might postulate that our negative results for BLV in human mammary tissue may be due to a low prevalence in cattle, analysis of USDA data show a high prevalence of BLV in northeast dairy herds, with 87% of herds infected and 40% of milk cows infected [88,89].Based on milk samples tested for anti-BLV antibodies by ELISA, the prevalence of BLV in dairy herds in Vermont was approximately 42% and this study did not find a significant association between standardized prevalence of BLV and herd size.[90] These numbers suggest that Vermont likely does not have low BLV infection rates relative to other regions, though this could potentially vary based on the size of the herds.Sequence variations in the BLV LTR region and tax gene are associated with promoter activity [78], which could affect viral loads and transmissibility in BLV-infected cattle at a regional level, but it is unlikely that such these small variations could preclude zoonotic transmissions in different regions of the U.S.
Our results are concordant with other negative findings in the literature that conclude that there is no apparent association of human breast cancers with BLV infection.As noted, Zhang et al. reported an inability to detect BLV in breast cancer tissue among 91 tissue samples from Chinese women whereas 50% of cattle blood samples had detectable BLV by PCR [68].Gillet et al. assessed whole genome sequences of 51 breast cancer tissue samples along with adjacent mammary tissue obtained from the NCBI Database of Genotype and Phenotype (dbGaP).Whole genome sequencing did not find any pairing with five different BLV variants [66].Whole genome analysis was performed to exclude "clonal integration of natural and highly divergent BLV strains in breast tumors."Their findings did not support an association between BLV and breast cancer [66].A recent study on 238 samples from the Alabama Hereditary Cancer Cohort FRET PCR and whole genome sequencing to detect the presence of BLV infection in association with human breast cancer, and no BLV DNA was detected [67].Likewise, our analysis does not support an association of BLV infection or the presence of BLV DNA or highly conserved BLV sequences with human breast cancers.
While, studies have shown animals aside from cattle to be host for BLV, including livestock such as sheep and buffalo [91], BLV has been absent from other animals, such as camels [92,93].Therefore, the ability of BLV to infect humans does require scrutiny.There are well-characterized biological mechanisms that preclude cross-species virus transmission, mechanisms originally identified in retroviral systems.These include a variety of restriction factors that effectively block virus replication in cells even from closely related species, such as human and non-human primates [94].There are additional biological constraints to human BLV infection.BLV is highly cell-associated; that is, infection requires the transfer of infected B lymphocytes from one host to another.Natural infections of cattle and water buffalo require parenteral transfer of blood by biting insects, cross contamination during veterinary procedures, or vertical lymphocyte transfer during parturition or in milk.Experimental infections of alternate hosts requires parenteral transfer of sufficient numbers of infected bovine lymphocytes from lymphocytic animals or high doses of cell-free virus.The infected lymphocytes and cell-free virus particles are sensitive to heat and desiccation.Pasteurization would inactivate infectious BLV in milk for human consumption, and extensive epidemiological studies have shown no association between milk consumption and leukemia in people drinking raw milk from infected cattle [95].In addition to innate cellular restrictions and required modes of transmission, there is the question of cell tropism.BLV tropism for B cells is well described with potential implications for other leukocyte populations [96].However, the ability of BLV to infect epithelial cells rather than lymphocytes upon zoonotic infection and the induction of carcinoma rather than lymphoma would require divergent receptors on the surfaces of these cell types in these mammalian species as the viral glycoproteins are largely unchanged.These hypothetical factors do not preclude the possibility of BLV induction of human breast cancer.However, these mechanisms are consistent with data sets such as ours that do not find a link between BLV DNA and human breast cancer as we see with the enzootic BLV infection and its resultant malignancy in cattle.

Conclusion
Our study does not support a high prevalence of BLV in breast tumor tissue, decreasing support for a potential association of BLV with breast cancer.Given contradictory studies of what would be a preventable cause of breast cancer, a cross-institutional study applying a variety of techniques to allow for sufficient confirmatory testing would be ideal to settle discrepant findings on the prevalence of BLV DNA in mammary tissue and its association with breast cancer.

Table 1
Primers and annealing conditions used for nPCR RoleGAPDH Fwd Human GAPDH GAG TCA ACG GAT TTG GTCGT 50Used to confer extracted human DNA quality GAPDH Rev TTG ATT TTG GAG GGA TCT CG tax Fwd 1 BLV tax CTT CGG GAT CCA TTA CCT GA 55 Used as outer primer pair for tax nPCR Amplicon A, and Tax nPCR Amplicon C tax Rev 1 GCT CGA AGG GGG AAA GTG AA tax Fwd 2 BLV tax ATG TCA CCA TCG ATG CCT GG 55 Used as outer primer pair for tax nPCR Amplicon B, and inner primer pair for tax nPCR Amplicon C tax Rev 2 CAT CGG CGG TCC AGT TGA TA tax Fwd 3 BLV tax GGC CCC ACT CTC TAC ATG C 56 Used as inner primer pair for tax nPCR Amplicon A, and Tax nPCR Amplicon B tax Rev 3 AGA CAT GCA GTC GAG GGA AC LTR Fwd 1 BLV LTR TAG GAG CCG CCA CCGC 57 Used as outer primer pair for LTR nPCR Amplicon LTR Rev 1 GCG GTG GTC TCA GCCGA LTR Fwd 2 BLV LTR AAA CTG CAG CGT AAA CCA GAC AGA GACG 58 Used as outer primer pair for LTR nPCR Amplicon LTR Rev 2 CAC CCT CCA AAC CGT GCT TG

Table 2
Patient characteristics