Abstract
Objective This retroactive meta-analysis assesses the relationship between Mycobacterium avium subspecies paratuberculosis (MAP) and Crohn’s Disease (CD).
Methods A meta-analysis of PCR– and culture–based studies was conducted to determine if there is a relationship between MAP and Crohn’s Disease. The chi-squared test of independence was also conducted to determine if MAP infection and Crohn’s disease onset are independent events.
Results The studies analyzed were able to provide evidence that Mycobacterium avium subspecies paratuberculosis is highly associated with Crohn’s disease. It is also shown that, regardless of detection method, MAP can effectively be detected in Crohn’s patients.
Conclusion MAP is strongly connected with Crohn’s Disease.
Summary This paper examines the incidence of MAP in the tissue of Crohn’s disease patients. We found evidence to support MAP being strongly associated with Crohn’s disease.
1. Introduction
Crohn’s Disease (CD) is an inflammatory bowel disease (IBD) of unknown cause. The disease can affect any part of the gastrointestinal (GI) tract but usually predominates in the small intestine [1]. Patients with Crohn’s often suffer from abdominal pain, excessive diarrhea, gas, weight loss, and other symptoms which can adversely affect their quality of life [2].
As of yet, there is no agreed-upon cause of the disease, although there is a wealth of various theories regarding its etiology. Theories range anywhere from autoimmunity [3] to Western dieting [4] to bacterial infection [5].
In the animal kingdom, there is an IBD whose disease progression is a near-perfect match with CD – this disease is known as Johne’s Disease (JD) [6]. JD affects many animals including monkeys, chimpanzees, and other primates, deer, cattle, sheep, and others. The causative agent of this disease is a bacterium known as Mycobacterium avium subspecies paratuberculosis (MAP).
Due to the similarities between JD and CD, it has been proposed by many researchers that MAP causes CD in humans as it does JD in animals [7,8].
The major issue with proving a causal relationship between the two is that it is very difficult to detect MAP using the more common microbiological techniques. There have been many unsuccessful attempts to isolate and culture MAP from Crohn’s tissue; MAP recovers very poorly and its incubation times are extremely long. Because of this hardship, very few researchers have been able to prove Koch’s Postulates for MAP and Crohn’s. In this era of post-genomics, however, researchers have been able to detect MAP via PCR; they can use PCR to detect a region in the bacterial genome known as IS900.
This study aims to use mathematical/empirical methods to determine how strongly associated is MAP with Crohn’s Disease by performing a retrospective meta-analysis on the studies that have used PCR and culture techniques to attempt to isolate MAP from adult Crohn’s patients.
2. Materials and Methods
2.1. Search Strategy
As there have already been papers published highlighting the studies that have investigated culture and PCR attempts to isolate MAP from Crohn’s tissue, this study did not rehash those same search strategies. Instead, to conduct this study, we took the studies reported in Naser et al’s paper [9], and meta-analyzed those studies. The studies were extracted from PubMed. No studies were excluded.
2.2. Data Extraction
Data were extracted onto a Microsoft Excel spreadsheet. First author, year of publication, number of participants in Crohn’s disease and control groups, exposure parameters (i.e., MAP infection/isolation), and event outcomes in both groups were obtained from each study.
2.3. Meta-Analysis
Probability difference (PD) was used as the measure of risk differences, as described by Whitehead, 2002 [10], and Log-odds ratio (LORs) were calculated using the Mantel-Haenszel method, as described by Whitehead, 2002 [10].
2.4. Mathematically determining association
To determine how strongly associated are MAP and Crohn’s disease, a chi-squared test of independence was utilized on the data, with the help of a 2×2 table, shown in Table 1 below.
2.5. Statistical Analysis
Statistical analysis was carried out using R Studio, R version 4.0.5 (2021-03-31) – “Shake and Throw”, Platform: x86_64-w64-mingw32/x64 (64-bit). All calculations were verified by hand using Microsoft Excel.
3. Results
3.1. Search Results
The studies used in this study were extracted from Naser et al’s (2014) [9] review on MAP in Crohn’s patients. Fifty-four papers [8,12-65] yielding 75 studies were identified; none were excluded. To verify these papers, Naser’s paper was cross-referenced with Barukčić’s (2018) [11] quasi-meta-analysis on the same topic; Barukčić’s used fewer studies without reasonable reasons. At any rate, his paper did not yield any additional studies that Naser did not cover. A search was then conducted on PubMed using the keywords of “Mycobacterium avium subspecies paratuberculosis” and “MAP” with “Crohn’s disease”, “IBD”, “PCR”, “DNA”, “Culture”, “IS900 PCR detection.” No additional papers were feasible to use, so Naser’s original 54 papers were used [8,12-65], yielding a total of 5323 patients. 23 studies investigated culturing MAP with a total of 1795 patients. Meanwhile, 48 papers yielding 53 studies investigated PCR detection of MAP DNA, with a total of 3528 patients.
3.2. Detection of MAP via culture
Mycobacterium avium subspecies paratuberculosis was attempted to be cultured from a total of 1,795 patients. Of that number, 820 had Crohn’s disease with 975 controls. In the Crohn’s group, MAP was able to be cultured from 196 patients (31.41%); this was possible with just 46 (4.95%) in the control group (Table 2).
A random-effects model meta-analysis measuring log-odds ratios was conducted on the data (Figure 1). The log-odds of MAP being detected in tissue from patients with Crohn’s disease versus control was -1.50 (95% CI: -2.10 – -0.90). This value denotes a strong association between Crohn’s disease status and MAP recovery. This translates to an 81.76% chance of recovering and culturing MAP from Crohn’s tissue.
Log-odds ratios (LOR) are indicated by the numbers on the x-axis. Weight: the percentage contribution of an individual study to the overall estimate. The vertical dashed line indicates the random effects log-odds ratio. The solid horizontal lines show the 95% confidence intervals (CI). The size of the black box and diamond is proportional to the corresponding weight.
A random-effects model meta-analysis measuring probability difference was also conducted. The probability difference (risk difference) was -0.23 (95% CI: -0.32 – -0.13).
A χ2-test of independence was also run on the data. The null hypothesis was that MAP infection and Crohn’s disease development are independent, with the alternative hypothesis being that MAP infection and Crohn’s disease development are not independent. The independence test yielded a χ2 value of 138.9 (p < 0.001), giving us enough confidence to be able to reject the null hypothesis, implying, at least in this case, that MAP infection and Crohn’s disease development are not independent events.
To further explore this point, we also calculated the probability that Crohn’s disease development and MAP infection are not independent events (PMAP=CD). PMAP=CD was 0.6267.
3.3. Detection of MAP IS900 DNA via PCR
The DNA of the IS900 region of Mycobacterium avium subspecies paratuberculosis was attempted to be detected via the usage of PCR in a total of 3,528 patients. Of that number, 2012 had Crohn’s disease with 1516 controls. In the Crohn’s group, MAP DNA was detected in 580 patients (28.83%); this was possible with just 197 (12.99%) in the control group (Table 3).
A random-effects model meta-analysis measuring log-odds ratios was conducted on the data (Figure 2). The log-odds of MAP DNA being detected in patients with Crohn’s disease versus control was -1.08 (95% CI: -1.51 – -0.64). This value denotes a strong association between Crohn’s disease status and MAP detection. This translates to a 74.65% chance of detecting MAP via PCR in Crohn’s patients.
A random-effects model meta-analysis measuring probability difference was also conducted. The probability difference (risk difference) was -0.16 (95% CI: -0.21 – -0.10). A χ2-test of independence was also run on the data. The null hypothesis was that MAP infection and Crohn’s disease development are independent, with the alternative hypothesis being that MAP infection and Crohn’s disease development are not independent. The independence test yielded a χ2 value of 125.27 (p < 0.001), giving us enough confidence to be able to reject the null hypothesis, implying, at least in this case, that MAP infection and Crohn’s disease development are not independent events. PMAP=CD was calculated as well, yielding a value of 0.5383.
3.4. Overall detection of MAP
Putting together all the studies, irrespective of detection type, Mycobacterium avium subspecies paratuberculosis was attempted to be detected from a total of 5,323 patients. Of that number, 2832 had Crohn’s disease with 2491 controls. In the Crohn’s group, MAP was detected in 776 patients (27.40%); this was possible with just 243 (9.76%) in the control group (Table 4). A random-effects model meta-analysis measuring log-odds ratios was conducted on the data (Figure 3). The log-odds of MAP being detected in tissue from patients with Crohn’s disease versus control was -1.21 (95% CI: -1.56 – -0.86). This value denotes a strong association between Crohn’s disease status and MAP detection. This translates to a 77.03% chance of detecting MAP in Crohn’s tissue.
A random-effects model meta-analysis measuring probability difference was also conducted. The probability difference (risk difference) was -0.18 (95% CI: -0.22 – -0.13). The independence test yielded a χ2 value of 265.47 (p < 0.001), giving us enough confidence to be able to reject the null hypothesis, implying that MAP infection and Crohn’s disease development are not independent events. Meanwhile, PMAP=CD was 0.5681.
4. Discussion
Crohn’s disease is a disease of the gastrointestinal tract, characterized by severe abominable pain, diarrhea, inflammation, and a host of other symptoms that cause patients to experience a very low quality of life for years with very little relief.
As was touched on briefly earlier, it has been shown that the cattle who exhibit symptoms of Johne’s disease do so in response to the bacterial agent Mycobacterium avium subspecies paratuberculosis, or MAP. In the case of the animals, there is a normal inflammatory response to an abnormal gut pathogen.
The ensuing host cellular immune response leads to the typical granulomatous enteritis pathognomonic of Johne’s and Crohn’s Diseases, characterized by the thick and corrugated appearance of the intestinal wall and inflamed lymph nodes.
Tissue macrophages and Dendritic Cells (DCs) play a crucial role in the recognition of pathogen-associated molecular patterns (PAMPs) in the innate phase via toll-like receptors (TLRs), as well as in antigen processing and the elicitation of cytokines-mediated cellular interactions. The bacteria, however, has evolved to escape phagocytosis by macrophages. Like its cousin Mycobacterium tuberculosis, MAP hides inside of the cell, while the macrophages – both infected and healthy – form granulomas that now harbor latent MAP infections. MAP is now free to replicate unchecked, feeding off of the body’s nutrients. Over time, MAP leaves the granuloma, causing inflammation afresh in the gastrointestinal tract – the same inflammation seen in Crohn’s Disease.
This study sought to provide mathematical evidence of MAP being associated with Crohn’s disease – whether as a causative agent or an opportunistic pathogen remains to be determined by researchers. First, a log-odds ratio was calculated for the two most common methods of MAP detection – culturing and PCR. Traditionally, it has been harder to detect the bacterium via culturing methods than it is when employing PCR; therefore, it is normally expected that any analysis using culturing methods would not yield any ‘good’ results showing a relationship between MAP infection and disease development. It would be expected that the results from the PCR analysis would show a greater correlation between the two factors. In this study, however, we found the Culture studies to be better proof positive.
The Culture Studies (21/75) yielded a total of 1795 patients. Of those, 820 had Crohn’s disease, while 975 were in the control group (the control group was defined as any patient who did not have Crohn’s disease specifically; this analysis was careful to differentiate the IBDs (Crohn’s, Ulcerative Colitis (UC), and Indeterminate Colitis (IC)) – if a person had UC or IC, they were classified as ‘control’ as we were investigating only the effect of MAP infection on Crohn’s patients specifically. This is important to note because many researchers falsely assert that because MAP is not positively associated with UC, MAP cannot be a cause of IBD, as if IBD were a lone disease in its own right. Crohn’s and UC are different diseases within the same class; therefore, it would stand to reason that they would have different causes while sharing some similar characteristics). In the Crohn’s group, MAP was recovered and cultured from 196 (23.9%) patients, while in the control group, MAP was recovered and cultured from only 46 (4.72%) patients. Perhaps more telling is the breakdown of MAP infection. The bacterium was able to be isolated and cultured from a total of 242 patients (13.5%). Of this group, a whopping 81% of these patients had Crohn’s disease! Meanwhile, 95% of patients who did not have MAP also did not have Crohn’s disease (929/975). These results garnered a χ2 value of 138.9 (p < 0.001); Pearson’s chi-squared test of independence with Yates’ continuity correction was employed.
These studies were also meta-analyzed with a random-effects model employed due to the high heterogeneity between studies (I2 = 44%) due to the unpredictable nature of actually being able to culture MAP in a laboratory setting. Of the 21 studies analyzed, 17 (81%) yielded a log-odds ratio (LOR) in favor of MAP infection being associated with Crohn’s disease development, with a total weight of 74.9%. The random-effects model LOR was -1.50 (95% CI: [-2.10 – -0.90]) (in this analysis, a LOR less than zero denoted correlation between MAP infection and Crohn’s disease development, with the opposite true for LORs above zero; a LOR value of zero denoted no specific association between the two variables). This LOR value roughly translates to a 0.7311 probability of MAP being recovered and cultured from any given patient who has Crohn’s disease.
The PCR Studies (54/75) yielded a total of 3528 patients. Of those, 2012 had Crohn’s disease, while 1516 were in the control group. In the Crohn’s group, the IS900 region of the MAP genome was detected in 580 (28.83%) patients, while in the control group, MAP DNA was detected in 197 (12.99%) patients. MAP was detected via PCR in a total of 777 patients (22.02%). Of this group, 74.65% of these patients had Crohn’s disease! Meanwhile, 87% of patients who did not have MAP also did not have Crohn’s disease (1319/1516). These results garnered a χ2 value of 125.27 (p < 0.001.
These studies were also meta-analyzed with a random-effects model. Of the 54 studies analyzed, 36 (66.67%) yielded a log-odds ratio (LOR) in favor of MAP infection leading to Crohn’s disease development, with a total weight of 69%. The random-effects model LOR was -1.08 (95% CI: [-1.51 – -0.64]). This LOR value roughly translates to a 0.7311 probability of MAP DNA being found in the tissue of any given patient who has Crohn’s disease.
When analyzing the studies as a whole (i.e., irrespective of MAP detection method), the 75 studies yielded a total of 5323 patients. Of those, 2832 had Crohn’s disease, while 2491 were in the control group. In the Crohn’s group, MAP was detected in 776 (27.4%) patients, while in the control group, MAP was detected in only 243 (9.75%) patients.
As a whole, MAP was detected in a total of 1019 patients (19.14%). Of this group, 76.15% of these patients had Crohn’s disease. Meanwhile, 90.24% of patients who did not have MAP also did not have Crohn’s disease (2248/2491). These results garnered a χ2 value of 265.47 (p < 0.001.
These studies were also meta-analyzed with a random-effects model (the bias was assessed via a funnel plot presented in Figure 4), giving a LOR of -1.21 (95% CI: [-1.56 – -0.86]). This LOR value roughly translates to the same probability of finding MAP as evaluating the two detection methods separately.
The dashed vertical line indicates the overall log-odds ratio, and the diagonal lines indicate the expected 95% confidence intervals associated with the mean LOR for studies with various numbers of study subjects.
When verifying the calculation of the chi-squared value by hand for the culture studies, we discovered that certain observations were higher than their expected counterparts. MAP recovery from Crohn’s patients was observed to be 77% higher than expected, while MAP recovery from those without Crohn’s was 65% lower than expected, when looking at a group of patients of the same sample size (1795). This shows that even though it is hard to culture MAP from tissue, MAP is still very much prevalent in patients with Crohn’s disease. There is no denying that Crohn’s disease and Mycobacterium avium subspecies paratuberculosis are, at the very least, related to each other.
The question remains: does MAP cause Crohn’s disease, or is MAP an opportunistic pathogen, taking advantage of an inflamed environment to invade, much like adherent invasive E coli? The answer to this very important question may very well change the way we think about, diagnose, and treat Crohn’s disease forever.
Data Availability
No databases were used; the data can be extracted from the papers analyzed (meta-analysis), listed in the references
Acknowledgements
We would like to acknowledge the Philadelphia AMP Bridge to Doctorate Program for supporting JP with the funding for his PhD program of which this present study is a part.
Footnotes
Conflict of Interest: The authors declare that they have no conflict-of-interest