Does mouse mammary tumor-like virus cause human breast cancer? Applying Bradford Hill criteria postulates

Background: The role of mouse mammary tumor-like virus (MMTV-like virus) in human breast cancer (BC) has already been widely investigated worldwide with conflicting results. Although the researchers tried to establish the link between MMTV-like virus and BC through the statistical meta-analysis of the previous studies associating MMTVlike virus with BC, they failed to establish a more reliable link due to the shortcomings of the statistical meta-analysis. In the present study, we identified population-wide studies relating MMTV-like virus with BC through the PubMed search engine. Then, we examined the available data of MMTV-like virus prevalence in BC, normal/benign samples, and applied the postulates of Bradford Hill criteria on the available evidence to investigate the association between MMTV-like virus and BC. In addition, to further enhance the reliability of our outcomes, we have also evaluated the methodologies of the previous studies to address the possibility of false-negative and false-positive results. Results: After a careful evaluation of the extracted data against the postulates of Bradford Hill criteria, it was observed that none of the studies fulfill all the major postulates of Bradford Hill criteria for causation including temporality, consistency, biological gradient, experiment, coherence, specificity, and analogy. Hence, no causal relationship has been suggested between MMTV-like virus and BC patients of the any included population. Conclusion: The results failed to prove the causal relationship between MMTV-like virus and BC rather suggested it as a co-participant in the pathogenesis of BC.


Background
Breast cancer (BC) is a condition in which the breast cells grow abnormally and out of control. A breast is composed of three main parts: lobules, ducts, and connective tissue. The lobules are the glands that produce milk, while ducts are used to carry the milk to the nipple, and connective tissue surrounds and holds everything together (Zucca-Matthes et al. 2016). BC may occur at multiple sites in the breast but ducts and lobules are the major hotspots of the BC. Based on the type of breast cells that turn into cancer, BC has been divided into several types (Sharma et al. 2010). BC is one of the leading causes of mortality worldwide accounting for 14% of cancer-related deaths (Jemal et al. 2011) with a 5-year survival rate of approximately 40-80% (Coleman et al. (2008)).
Being female, getting older, consuming alcohol, no pregnancy, obesity, and viral infection are the very common causes of the BC (Lawson 2009). Moreover, the role of human papilloma (HPV) and Epstein-Barr virus (EBV) in the development of BC has already been well acknowledged worldwide . There is, however, no clear etiological role of mouse mammary tumor viruses (MMTV) have been identified in the development of BC since the first study which initially identified the presence of MMTV in human BC in 1972 (Axel et al. 1972;Oskouee et al. 2014).
Another virus named MMTV-like virus is 95% homologous to MMTV in terms of sequence homology . Controversial evidence has been reported until now regarding the etiologic role of MMTV-like virus in the development of BC. Although the role of MMTV as an etiologic factor in the development of mice memory tumor is well established, however, it is still unclear whether MMTV-like virus contributes to the development of human BC.
The first-ever study documenting the presence of MMTV-like virus was conducted in 1972 by Axel et al. (1972) in the USA. In that study, they detected the presence of MMTV-like virus in a total of 38 BC samples using the molecular hybridization technique and results revealed the presence of MMTV-like virus in approximately 78.9% of the BC samples.
In general, a statistical meta-analysis is usually preferred when establishing a correlation between the virus and the disease as compared to the single study. This choice is based on the multiple advantages of the metaanalysis such as increased number of objects, greater diversity among the objects, and conclusion with a high level of evidence over the individual single study which has disadvantages like a small cohort of patients and conclusions with a low level of evidence. By keeping in view the inconsistencies in the MMTV-like virus detection ratios in worldwide published studies, recently, researchers have analyzed the previously published studies by the means of statistical meta-analysis to yield more useful pieces of information.
Previously, a statistical meta-analysis was performed to find out the causal relationship between MMTV-like virus and BC by Wang et al. (2014) through the available literature of MMTV-like virus and BC in various authentic research engines including MEDLINE (Pub-Med), Embase, EBSCO (ASP/BSP), and China National Knowledge Infrastructure (CNKI). They obtained more than 22 studies from different populations such as Western, Asian population and African populations and their revealed no significant association between MMTV-like virus and BC.
Although evaluating the results of previous studies documenting the role of MMTV-like virus in the development of BC through statistical meta-analysis was a better choice than generalizing the results of an individual study, however, we did not consider statistical meta-analysis reliable to establish a causal-relationship between MMTV-like virus and the BC development because of some serious limitations such as its inability to analyze the methodologies of the previous studies, so there is no way to evaluate the possibility of false-negative and false-positive results, nor does statistical meta-analysis provide any information regarding the effect of heterogeneity-specific nature of the understudied populations on MMTV-like virus detection. In addition, statistical metaanalysis results in publication biasness, where meta-analysis does not select studies with no results even though they contain valuable information.
By looking at the discrepancy in the outcomes of the previously published studies and significant shortcomings of the statistical meta-analysis, we performed the population-wise valuation of the results of previous studies detecting MMTV-like virus in BC using the Bradford Hill criteria. These criteria are widely used and accepted worldwide over many years for establishing a causal relationship between a presumed cause and an observed effect of public health research (Fedak et al. 2015).
In the course of evaluation, we analyzed, whether or not these studies fulfill all the postulates of Bradford Hill criteria to declare a causal relationship between MMTVlike virus and BC. In addition, we also evaluated the methodologies used by the previous studies to address the possibility of false-positive and false-negative results for better outcomes. The outcomes of the present study will help to establish a more reliable population-wise causal relationship between MMTV-like virus and BC and determine the more appropriate treatment strategies for BC patients.

Methods
In the present study, we implemented a two-phase methodology ( Fig. 1).

Literature search
All the relevant articles associating MMTV-like virus with BC were identified through the PubMed search database using the keywords: "Breast Cancer" AND "MMTV-like virus. " We also defined "Retroviridae" AND "Breast neoplasia" as medical subject headings (MeSH) terms. Mesh terms and keywords were combined during the search process. All the literature was searched available until March 2020, with the "Original Article" filter. In total, 1428 original articles were identified through the PubMed search engine.

Relevant data extraction
From 1429 original articles, the 40 relevant articles were identified having the desired information by initially Page 3 of 13 Hameed et al. Bull Natl Res Cent (2020) 44:183 reading the title, abstract, and then the complete article. Furthermore, a comprehensive table was constructed having all the required information from the selected relevant studies.

Evaluation of the results using the postulates of Bradford Hill criteria
Based on the extracted data, all the identified studies were carefully evaluated against the following Bradford Hill criteria postulates: (1) Strength: larger the association, more probability of the causal relationship, (2) Temporality: cause must lead to the induction of an effect. If the delay is expected between the cause and effect, then the effect has to occur after the delay, (3) Consistency: different studies conducted by different researchers at different places with different sample sizes and reporting the similar results increase the chances of the causal relation between the cause and effect, (4) Plausibility: there should be plausible mechanism between the cause and effect, (5) Biological gradient: greater response is produced by the causative agent in response to the greater exposure. However, in some cases, effect can be triggered by the mere presence of the factor, while in other cases, greater exposure can lead lower effect as well, (6) Experiment: the relationship between the cause and effect should be explained by the experiments and experiment should results in the reduction of effect when the causative agent is removed, (7) Coherence: causal relationship should not conflict with already known literature about the disease or exposure, (8) Specificity: causality is more likely if the effect has only one cause, (9) Analogy: previous evidence of the association between the cause and effect should support the current statement for the causal relationship.
The assessment of each postulate was qualitative/ descriptive, as there was an element of subjectivity in applying quantitative scoring. Evidence collected for each postulate is presented in Table 1 and results section with a final judgment as to whether the viewpoint was fulfilled or not. Page 5 of 13 Hameed et al. Bull Natl Res Cent (2020)   Page 6 of 13 Hameed et al. Bull Natl Res Cent (2020)
The results obtained after careful evaluation of the extracted data through the Bradford Hill criteria postulates showed that all the identified studies from various populations do not fulfill the major postulates including temporality, consistency, biological gradient, experiment, coherence, specificity, and analogy. Hence, no causal relationship has been suggested between MMTV-like virus and BC patients of any included population.
In Pakistan, the association between MMTV-like virus and BC has been reported in only 03 studies  so far. They utilized PCR technique with primers specifically targeting the Env and LTR regions of the viral genome and documented varying MMTV-like virus detection positivity ratio ranging from 11.6%  to 46.25%  in BC samples.
A total of 05 studies Motamedifar et al. 2012;Tabriz et al. 2013;Reza et al. 2015;Shariatpanahi et al. 2017) including 04 case-control studies Motamedifar et al. 2012;Reza et al. 2015;Shariatpanahi et al. 2017) have been carried out in Iran so far, analyzing the association between MMTV-like virus and BC. These studies utilized PCR technique with primers specific for the Env region of the viral genome and documented varying MMTV-like virus detection positivity ratios ranging from 0% Motamedifar et al. 2012;Tabriz et al. 2013) to 32.2%  in BC samples, while 0% Motamedifar et al. 2012;Reza et al. 2015) and 5%  in the normal and adjacent/benign controls, respectively.
A single case-control study ) has been conducted in Morocco so far to determine the association between MMTV-like virus and BC. They analyzed 42 BC samples and 18 adjacent/benign controls for the identification of MMTV-like virus in BC using PCR technique with primers specifically hybridizing in the Env sequence of the viral genome. They documented 57.1% and 33.3% MMTV-like virus detection positivity ratios in BC and adjacent/benign controls, respectively.
In Austria, only a single study  has been reported to date to relating MMTV-like virus presence with BC. They screened 50 BC samples using PCR technique with primers specifically targeting the Env sequence of the viral genome and they documented 0% MMTV-like virus detection positivity ratio in BC samples.
Until now, only a single case-control study ) has been carried out in Egypt to find out the etiological association between MMTV-like virus and BC. They analyzed 100 BC samples and 50 adjacent/benign controls using PCR technique with Env gene-specific primers for MMTV-like virus detection. They reported the higher MMTV-like virus detection positivity ratio in BC samples (36%) as compared to the adjacent/benign controls (4%).
In Vietnam, there has been a single case-control study  reported so far to find out if MMTVlike virus has any association with BC. They utilized PCR technique with primers specifically targeting the Env sequence of the viral genome and documented a 0.83% MMTV-like virus detection positivity ratio in BC samples and normal controls, respectively.
Until now, only a single case-control study  has been carried out in Argentina to elaborate the role of MMTV-like virus in the development of BC. They screened 74 BC samples paired with adjacent/ benign controls using PCR technique with primers targeting the envelope sequence of the viral genome. They documented the higher MMTV-like virus detection positivity ratios in BC samples (31%) as compared to the adjacent/benign samples (0%) and normal controls (1.35%).
In Sweden, a single case-control ) has been carried out so far to identify the casual relationship between MMTV-like virus and BC using RT-PCR technique and they did not detect MMTV-like virus in any of the BC or control sample.
So far, a single case-control study ) has been carried out in Saudi Arabia to find out the association between MMTV-like virus and BC. They screened a total of 101 BC, 51 normal, and 93 adjacent/ benign controls using PCR technique with primers specific for the Env region of the viral genome and documented MMTV-like virus positivity ratios was higher in the adjacent/benign controls (9.7%) as compared to the normal (0%) and BC samples (5.9%).
In Tunisia, a total of 03 studies Levine et al. 1984;Pogo et al. 2010) including 02 casecontrol studies Pogo et al. 2010) have been reported so far to find out whether MMTVlike virus have any association with BC. They utilized antigen detection method and PCR technique with Env gene-specific primers for MMTV-like virus detection and documented MMTV-like virus detection positivity ratios with varying frequencies ranging from 3.9% Page 10 of 13 Hameed et al. Bull Natl Res Cent (2020) 44:183  to 73.7%  in BC samples, while 0% ) and 1.4%  in adjacent/benign and normal samples, respectively.
To date, a total of 04 studies (Axel et al. 1972;Etkind et al. 2000;Pogo et al. 2010;Wang et al. 1995) including 3 case-control studies Pogo et al. 2010;Wang et al. 1995) have been carried out in the USA to elaborate the causal association between in adjacent/benign samples and BC. All these studies employed PCR technique with Env gene-specific primers and documented MMTV-like virus detection positivity ratios ranging from 36.9%  to 45.5%  in BC samples, while 6.9%  in the adjacent/benign samples and 0%  to 1.8%   A single study ) has been conducted in Myanmar to date to relate the MMTV-like virus presence with BC. In total, they analyzed 58 BC samples were using PCR technique with primers specific for the Env region and documented a 1.7% 1.7% detection positivity ratio.
A total of 04 studies Naccarato et al. 2019;Pogo et al. 1999;Zammarchi et al. 2006) including a case-control study  have been carried out in Italy so far, analyzing the etiological association between 1.7% and BC. These studies utilized the PCR technique with primers specific for the Env region of the viral genome and documented MMTVlike virus detection positivity ratios with varying frequencies ranging from 33%  to 68.1%  in BC samples, while 0%  in normal and 19.2%  in adjacent/benign controls.
In China, there has been a single study  reported so far to find out if MMTV-like virus is associated with BC. They utilized PCR technique with primers specifically targeting the Env sequence of the viral genome and documented 18.25% MMTV-like virus detection positivity ratio.

Discussion
BC is one of the most common types of cancer that infect millions of people worldwide each year. Although recent advancements in the diagnosis and treatment of the BC have helped to manage the disease but still the prevalence of BC is on a rise due to unknown underlying mechanisms (Akram et al. 2017).
To date, various individual studies have been carried out worldwide to find the state of association between MMTV-like virus and BC to further uncover the molecular pathways underlying BC but their results are conflicting. In addition, the statistical meta-analysis was also used by the researchers to analyze the previous individual studies for generating a more meaningful association between MMTV-like virus and BC but due to the shortcomings of statistical meta-analysis, researchers once again failed to establish a more reliable causal relationship between MMTV-like virus and BC.
In the present study, we evaluated the previous studies using a reliable, Bradford Hill criteria to find a causal relationship between MMTV-like virus and BC. In addition, we also evaluated the methodologies used by the previous studies to address the possibility of false-positive and false-negative results for better outcomes.