key: cord-0300771-ep8v0yih
authors: Blohs, Marcus; Mahnert, Alexander; Brunnader, Kevin; Flucher, Christina; Castellani, Christoph; Till, Holger; Singer, Georg; Moissl-Eichinger, Christine
title: Acute appendicitis manifests as two microbiome state types with oral pathogens influencing severity
date: 2022-04-14
journal: bioRxiv
DOI: 10.1101/2022.04.13.488268
sha: 3b48baf0a15a133d68c7b8cf153c6b1570627bfd
doc_id: 300771
cord_uid: ep8v0yih

Mounting evidence suggests that acute appendicitis (AA) is not one but two diseases: complicated appendicitis, which is associated with necrosis leading to perforation or periappendicular abscess, and uncomplicated appendicitis, which does not necessarily result in perforation. Even though AA is the most frequent cause of surgery from abdominal pain, little is known about the origins and etiopathogenesis of this disease, much less regarding the different disease types. In this study, we investigated the microbiome of samples from the appendix, rectum and peritoneum of 60 children and adolescents with AA to assess the composition and potential function of bacteria, archaea and fungi. The analysis of the appendix microbial community revealed a shift depending on the severity of the AA. This shift was reflected by two major community state types that represented the complicated and uncomplicated cases. We could demonstrate that complicated, but not uncomplicated, appendicitis is associated with a significant local expansion of oral, bacterial pathogens in the appendix, most strongly influenced by necrotizing Fusobacterium spp., Porphyromonas and Parvimonas. Uncomplicated appendicitis, however, was characterised by gut-associated microbiomes. Our findings support the hypothesis that two disease types exist in AA, which cannot be distinguished beyond doubt using standard clinical characterization methods or by analysis of the patient’s rectal microbiome. An advanced microbiome diagnosis, however, could improve non-surgical treatment of uncomplicated AA. Importance With a lifetime risk of up to 17%, acute appendicitis is one of the most frequent causes of emergency abdominal surgery in westernized countries. Latest literature reports suggests that appendicitis manifests in two disease types: complicated and uncomplicated appendicitis with different, yet unknown, etiopathogenesis. In this study, we investigated the microbial composition (bacteria, archaea and fungi) from 60 children and adolescents that were diagnosed with acute appendicitis. Appendix, rectal and peritoneal samples were analysed using amplicon and metagenomic sequencing. Our results suggest that acute appendicitis manifests in three microbial state types that reflect complicated and uncomplicated appendicitis as well as special cases that are caused by bacterial overgrowth. Strikingly, uncomplicated appendicitis appears to be caused by gut-associated pathogens while complicated appendicitis is driven by oral-associated microbes such as Fusobacterium sp. or Porphyromonas sp. The findings provided in our study are of special interest to understand the etiopathogenesis of both complicated and uncomplicated appendicitis.

Introduction appendix and peritoneum samples. Fusobacterium has been repeatedly reported to be 158 associated with disease severity in AA (12, 13, 15, 20, 23, 24) , and the results of differential 159 abundance analysis confirm this finding in our dataset (Fig. 2D ). By applying MaAsLin2 160 (Multivariable Association Discovery in Population-scale Meta-omics Studies), we identified 161 a total of nine differentially abundant genera, in which the abundances significantly 162 increased or decreased upon disease progression in appendix samples. In addition to 163 observing a significant increase in the relative abundance of Fusobacterium signatures, we 164 also observed an increased abundance of other typically oral cavity-associated microbes 165

including Parvimonas, Peptostreptococcus and Solobacterium. Strikingly, all these 166 (potentially) opportunistic pathogens show a very low abundance at catarrhal severity (< 1% 167 fact, all of the perforated cases cluster together. This CST is postulated to represent cases 213 of complicated appendicitis, with the key taxa including oral-cavity-associated 214

Fusobacterium, Porphyromonas and Parvimonas species (Fig. 2C ) which, as we and others 215

show (12, 13, 15, 23, 24) , correlate closely with disease severity. CST 3, on the other hand, 216 is believed to represent cases of uncomplicated appendicitis, as this group contains almost 217 all cases of catarrhal appendicitis, with the gut-associated Bacteroides and Faecalibacterium 218 as the dominant taxa. The major discriminating factor between these clusters is the disease 219 severity, which is significantly higher in CST 2 as compared to CST 3 (chi-squared test, P = 220

.004). This is also reflected in the number of patients with elevated blood leukocytes: 86.3% 221 of CST 2 patients were diagnosed with leucocytosis but only 57.1% of CST 3 patients (P = 222 .048). However, other diagnostic parameters did not show significant associations with both 223

CSTs, including CRP (Wilcoxon rank-sum test; P = .470), PAS (P = .077) and Alvarado 224 score (P = .083). Interestingly, we also observed another cluster of four samples in CST 1. 225

This group is unique as we identified very high abundances of either Haemophilus (rel. 226 abundance of 98.9% and 40.7%) or Escherichia-Shigella (rel. abundance of 28.9% and 227 53.5%) in the corresponding samples. CST 1 thus represents the "bacterial overgrowth" 228 cluster, in which a single genus or species is hypothesized as being responsible for the AA. Porphyromonas species (P. uenonis, P = .006 and P. asaccharolytica, P = .01) were also 240 enriched in gangrenous/perforated samples. However, none of the enriched taxa were found 241 to be significantly differently abundant after FDR correction. Such a lack of significant 242 differences was also reported recently by Yuan et al. (25) who used a similar classification 243 for AA severity. 244

Again, it is likely that pathological categorisation is not necessarily linked to the microbial 245 composition alone, especially for phlegmonous cases. Therefore, we applied the 16S rRNA 246 gene amplicon sequencing-based CST clustering method to our metagenomic data and 247 performed a differential abundance analysis on both the taxonomic and functional levels. 248

Based on MaAsLin2, only a significant enrichment of Fusobacterium necrophorum (P adj = 249

.035) was determined in CST 2 as compared to CST 3, further highlighting the importance of 250

Fusobacteria with respect to disease severity. Porphyromonas asaccharolytica was also 251 found to be enriched in CST 2 but not significantly after FDR correction (P adj = .14). This 252 microbial shift was accompanied by an altered abundance of functional genes in the 253 community. In particular, we observed a higher abundance of catabolism pathways in CST 2 254 and especially for amino acids, including lysine fermentation to crotonyl-CoA (P adj = .035), 255 histidine degradation (P adj = .051), glutamate fermentation (P adj = .066) and the associated 256 Na-driven 2-hydroxyglutarate pathway (P adj = .051), as well as the bacterial proteasome 257 pathway (P adj = .051). 258

To further validate and test the robustness of the results reported above, DESeq2 was 259 performed (Fig. 3) . On the taxonomic level, F. necrophorum was confirmed as being 260 significantly enriched in more complicated cases (CST 2) but also F. nucleatum, two 261

Porphyromonas species (P. endodontalis and P. uenonis) and two unspecified species of 262 the genera Prevotella and Alloprevotella were significantly enriched in CST 2. Despite the 263 marked expansion of those species, no significant change was observed at the functional 264 level. As noted in the MaAsLin2 analysis, an enrichment of catabolic pathways was apparent 265 in CST 2, indicating a potentially increased release of nutrients by, e.g. apoptotic or necrotic 266 host cells. However, this hypothesis is highly speculative and needs to be verified via 267 physiological characterisation of the corresponding species. 268

Neither the differential abundance nor ABRicate analysis yielded any indications that 269 antimicrobial resistance or virulence genes were enriched among the disease severities. and Solobacterium. Likewise, we could provide support for the previous observation of a 282 stepwise decline in gut-associated Bacteroides as the disease severity increased (13, 20, 283 23, 24). This decline is accompanied by a substantial reduction in the relative abundance of 284

Ruminococcaceae, Collinsella and Coprococcus from catarrhal to phlegmonous and/or 285 gangrenous/perforated appendicitis (Fig. 2D) . 286

For over a decade now, researchers have reported that uncomplicated and complicated 287 appendicitis do not share the same etiopathogenesis. This claim is supported by the 288 observation -among others -that not all cases of AA lead eventually to perforation. While 289 the time span from pain onset to surgery positively correlates with perforation, some cases 290 remain phlegmonous even after a long duration of pain (61-63). Our data on the microbial 291 composition support this difference in etiopathogenesis for AA, as we observed a decisive 292 microbial shift from catarrhal to gangrenous/perforated appendicitis, indicated by the results 293 of the differential abundance and beta diversity analyses as well as by de novo CST analysis 294 ( Fig. 3) . We identified three CST clusters with different characteristics. The first cluster is 295 associated with bacterial overgrowth typified by either Haemophilus or Escherichia-Shigella 296 (CST 1). The second cluster was enriched with Fusobacteria and other oral cavity-297 associated microbes (such as Porphyromonas and Parvimonas), a characteristic which 298 appears to be a hallmark of complicated appendicitis (CST 2). In the third cluster, cases are 299 defined by higher relative abundance of typical gut-associated bacteria such as Bacteroides 300

and Faecalibacterium without an apparent enrichment of oral microbes. These cases might 301 be unlikely to develop perforation and can be attributed to uncomplicated AA. However, 302 while CST 3 and CST 2 clearly separate catarrhal and perforated appendicitis, the 303 phlegmonous and (to a minor extent) gangrenous cases are distributed in both clusters. It is 304 plausible that the microbial community is only partially responsible for AA severity and that 305 some patients may develop complicated appendectomy with a CST-3-like microbial 306 composition. AA is a multifactorial disease that depends on multiple aspects including 307 lifestyle, diet and genetic predisposition (1, 2, 4, 16). It is apparent that proper tissue function 308 and microbial homeostasis require a delicate balance between the immune system, is a mutualistic microorganism that interacts with human tissue in ways that range from 331 abundance of the catabolic pathways for the production of the former three amino acids 350 were enhanced in CST 2. Especially lysine fermentation to crotonyl-CoA was predominantly 351 found to be enriched in more severe cases (CST 2), and this pathway has been described in 352 only few microbes, including F. nucleatum and Porphyromonas gingivalis (71). It is tempting 353 to speculate that Fusobacteria actively trigger apoptosis in intestinal epithelial cells during 354 AA to release peptides and amino acids. Austria. In brief, SequalPrep™ normalisation plates (Invitrogen) were used to normalise the 498 DNA concentration, and each sample was subsequently indexed with a unique barcode 499 sequence (8 cycles index PCR). All indexed samples were pooled, and the products of the 500 indexing PCR were purified with gel electrophoresis. Sequencing was performed using an 501

Illumina MiSeq device and the MS-102-3003 MiSeq® Reagent Kit v3-600 cycles (2x251 502 cycles). 503

The MiSeq data for all three approaches (universal, archaeal, fungal) were analysed 504 individually using QIIME2 V2019.11 (32) as described previously (33). Briefly, the DADA2 505 algorithm (34) was used to demultiplex and de-noise truncated reads as well as to generate 506 amplicon sequence variants (ASVs). Taxonomic assignment was based on the SILVA v138 507 database (35) for prokaryotic specimens and on the UNITE v8.3 database (36) for fungi. 508

Fungal raw reads were also pre-processed with ITSxpress (37), trimming reads to the 509 desired ITS2 region. The datasets was filtered as follows: Potential contaminants were 510 identified and removed with the R software package decontam (38) by providing negative 511 controls (DNA extraction and PCR negative controls) and applying a threshold of 0.25. 512

Control samples were subsequently removed from the dataset. Unassigned ASVs, those 513 classified as chloroplast and mitochondria, and ASVs with fewer than 10 total reads were 514 also removed. Rarefaction of the datasets was performed by scaling with ranked 515 subsampling (SRS) (39) using rarefaction depths of 1000, 100 and 50 for bacteria, fungi and 516 archaea, respectively. test, Kruskal-Wallis tests were applied to nonparametric data, while the comparison of 552 parametric data was conducted with a one-way analysis of variance (ANOVA) and a Tukey 553 post-hoc test. Comparisons of categorical data were performed with the chi-square test. 554

Microbial data analysis and visualization were done with QIIME2 (V2019.11) and R (V4.1.1). 555

Phylogenetic distances for the amplicon sequencing data were calculated in QIIME2 using 556 the fasttree plugin (52) and subsequently analysed following the core-metrics-phylogenetic 557 command without further subsampling. Biplots were performed based on the weighted 558 UniFrac distance matrix and were calculated using the biplot plugin in QIIME2. The Emperor 559 plugin (53, 54) was used to illustrate the results of the biplot analysis. 560 Differential abundance and alpha diversity testing as well as visualisation were performed 561 using normalised data within R and the packages phyloseq (55), ggplot2 (56), Maaslin2 (57), 562 DESeq2 (58). A full list of all packages and versions used can be found in Table S3 . Alpha 563 diversity was calculated based on the filtered and normalised dataset, and mean values ± 564 standard deviation (SD) are reported. After testing for normality, pairwise Wilcoxon signed-565 rank tests were performed, and the resulting P-values were corrected for false discovery 566 rates (FDR) according to Benjamini and Hochberg (59 PAS score (0-10) a ± SD 6.5 ± 1.9 6.6 ± 2.3 6.7 ± 2.3 8.7 ± 1.5 .19

Alvarado score (0-12) a ± SD 6.8 ± 1.9 6.6 ± 2.6 6.6 ± 2.4 8. 

The epidemiology of appendicitis 603 and appendectomy in the united states

The Epidemiology of Appendicitis

Appendectomy in South Korea: National Registry Data

A five-year longitudinal observational study in morbidity and 608 mortality of negative appendectomy in Sulaimani teaching Hospital/Kurdistan

Microbial composition of 631 human appendices from patients following appendectomy

Acute appendicitis in 633 children is associated with an abundance of bacteria from the phylum Fusobacteria

The fecal and mucosal microbiome in acute appendicitis 637 patients: an observational study

Acute 639 Appendicitis in Children Is Associated With a Local Expansion of Fusobacteria

Innate Immunity Genes Influence the Severity of Acute Children with and without Appendicitis

The oral microbiome-the relevant reservoir 659 for acute pediatric appendicitis?

Acute appendicitis is associated with 663 appendiceal microbiome changes including elevated Campylobacter jejuni levels

Exploring the Archaeome: Detection of Archaeal 685 Signatures in the Human Body

Comparison and validation of some ITS primer pairs useful for fungal 688 metabarcoding studies

High-resolution sample inference from Illumina amplicon data

The SILVA ribosomal RNA gene database project: improved data 716 processing and web-based tools

The UNITE database for molecular identification of fungi: handling dark taxa 720 and parallel taxonomic classifications

MEGAHIT: an ultra-fast single-741 node solution for large and complex metagenomics assembly via succinct de Bruijn 742 graph

MaxBin: an automated 744 binning method to recover individual genomes from metagenomes using an 745 expectation-maximization algorithm

Fast and sensitive protein alignment using

Functional Analysis in Metagenomics Using 749 MEGAN 6

Multivariable association discovery in population-scale meta-773 omics studies

Moderated estimation of fold change and 775 dispersion for RNA-seq data with DESeq2

Controlling the False Discovery Rate: A Practical and 777

Powerful Approach to Multiple Testing

Co-occurrence of anaerobic bacteria in colorectal

Analysis of Fusobacterium persistence and antibiotic response in colorectal cancer

Clinical value of total white blood cells and neutrophil counts in 826 patients with suspected appendicitis: retrospective study

The Decreasing Incidence of Acute 829

Zvizdic Z, Vranic S. 2021. Decreased number of acute appendicitis cases in pediatric 832 population during the COVID-19 pandemic: Any link?

Oral microbiome: possible harbinger for children's 835 health

For appendix samples (C), 858 the biplot analysis of the six most important ASVs (weighted UniFrac PCoA) and (D) the nine genera 859 that are significantly different between severity grades are shown

The catarrhal severity grade was used as a reference for the differential abundance analysis. FIGURE 3 | Community state type (CST) analysis for appendix samples. (A) shows a heatmap of the 863 25 most abundant genera in the appendix samples, sorted by the three defined CSTs

CST clustering is shown and based on Principal Coordinates Analysis (PCoA). (C) Indicates the most 866