Primary human mononuclear cells were isolated from tonsils obtained from patients (total n=54 taking into account all experiments, the particular number of samples per experiment were detailed in the corresponding Figure legends) undergoing tonsillectomy due to OSA. TMC were prepared as follows. Briefly, tonsils were collected in phosphate buffered saline (PBS) buffer containing 50 µg ml^-1 amphotericin B (Richet, BA, Arg). Tissues were chopped with a scalpel in IMDM medium (see below) and passed through a 70 µm-pore-size cell strainer (Falcon, Thermo Fisher, BA, Arg). TMC were purified by density gradient centrifugation with Ficoll-Hypaque (GE Healthcare, Uppsala, Sweden). The viability of primary cells, as determined by trypan blue exclusion was greater than 99% in all preparations. Informed consent was obtained from subjects before the study. The institutional ethics committee (Clinical Hospital, School of Medicine, Buenos Aires) approved the collection and use of clinical material, conformed to the provisions of the Declaration of Helsinki (as revised in Edinburgh 2000). Informed consent was obtained from all participants and/or their legal guardian/s. FACS experiments were performed with freshly isolated cells only.

TMC were cultured in IMDM medium (Life Technologies, CA, USA) containing 10% heat-inactivated fetal calf serum, 2mM L-glutamine, 100 U/ml penicillin, 100 µg/ml streptomycin, 20 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid buffer (HEPES), 1 mM sodium pyruvate and 50 µM 2-mercaptoethanol (all from Invitrogen, CA, USA). Human IL2 (20 ng/ml; R&D Systems, MN, USA) and human IL4 (20 ng/ml; R&D Systems, MN, USA) were added immediately before experiments also as supplements. When indicated, human recombinant hCD40L (250 ng/ml; R&D Systems, MN, USA) and 25 µM CpG-ODN 2006 (InvivoGen, CA, USA) were used. Cells were cultured at 1x10⁶ cells/ml either in 24-well culture plates (1ml) or 48-well culture plates (0.5ml).

Fluorochrome conjugated mAbs specific for human CD3 (Pacific Blue, clone UCHT1), human CD20 (FITC, clone L27 and APC H7 clone 2H7), human CD4 (PerCP, clone SK3), human IL17A (APC, clone BL168), TNFα (PE, clone IT-5H2 and BV711, clone Mab11), and respective isotype control mAbs were purchased from BD Biosciences (CA, USA) and Biolegend (CA, USA). Fixable viability dyes used were either eFluor 780 (eBioscience, CA, USA), Zombie Green (Biolegend, CA, USA) or Zombie Aqua (Biolegend, CA, USA) depending on the staining scheme. To detect intracellular cytokines in cultured cells, the latter were stimulated with PMA/Ionomycin/Brefeldin A for the last 5 hs of culture. Then, they were incubated with Cytofix/Cytoperm (BD PharMingen, CA, USA) for 20 minutes (min) in the dark and washed with Perm/Wash solution (BD PharMingen, CA, USA). Following permeabilization, the cells were stained with the respective anti-cytokine mAb. Cells were acquired using FACSAria II (BD Biosciences, CA, USA) and analyzed with FlowJo software (Treestar, OR, USA). Single stained controls were used to set compensation parameters. Fluorescence minus one and isotype-matched Ab controls were used to set analysis gates.

The results were analyzed using GraphPad Prism 7.0 software (Graph Pad Inc, CA, USA). The normality of variable distribution was assessed by the Shapiro–Wilk test, the hypothesis of normality was accepted when p > 0.05. Once the hypothesis of normality was accepted, comparisons were performed using unpaired Student t test. When normality resulted rejected, a Mann-Whitney test was used. In both cases, the null hypothesis being that 2 populations do not differ.

Tonsillar samples were cultured on Columbia agar containing 5% sheep blood and chocolate agar (Laboratorios Britania, Argentina) at 37°C with 5% CO2 for 24 to 48 h. Anaerobes were cultured onto Brucella agar supplemented with hemin and vitamin K under anaerobic conditions. All isolates were identified by MALDI-TOF MS. The isolates were identified by the direct colony on plate extraction method as previously described (8). Mass spectra were acquired using the MALDI-TOF MS spectrometer in a linear positive mode (Microflex, Bruker Daltonics). Mass spectra were analyzed in a m/z range of 2,000 to 20,000. The MALDI Biotyper library version 3.0 and MALDI Biotyper software version 3.1 were used for bacterial identification.

Cryostat sections (5-10 µm thickness) of tonsils were fixed and stained with mouse anti-human CD1d (clone NOR3.2/13.17 Santa Cruz Biotechnology, CA, USA) followed by chicken anti-mouse IgG antibody AlexaFluor 488 (Thermo Fisher Scientific, BA, Arg). (IgG/IgM)⁺ cells were detected by addition of goat anti-human IgM+IgG (H+L) F(ab´)₂ (Jackson Immunoresearch, PA, USA) followed by donkey anti-goat IgG AlexaFluor 594 (Jackson Immunoresearch, PA, USA). IgA⁺ cells were detected by addition of goat anti-human IgA F(ab´)₂ (InVivoGen, CA, USA) followed by anti-goat IgG AlexaFluor 594 (Jackson Immunoresearch, PA, USA). Cell nuclei were visualized with 4,6-diamidino-2-phenylindole staining (DAPI, Thermo Fisher Scientific, BA, Arg). Finally, slides were rinsed with phosphate buffered saline, air dried, mounted in Vectashield (Vector Labs, CA, USA) and sealed with a glass coverslip. Samples were examined with a Nikon Eclipse Ti-E fluorescence microscope (Nikon instruments Inc, Tokyo, Japan). Corresponding isotype controls were always added (see Supplementary Information ).

A solution of 0.5 mg/ml of lysozyme (Sigma Aldrich, Darmstadt, Germany) in 0.1 M Tris-HCl and 0.05 M Na2EDTA was added to the cryosections, followed by incubation at 37°C for 3 h. Fixed, permeabilized tonsillar sections were then incubated in a moist chamber for 4hs at 48°C in hybridization buffer [0.9 M NaCl, 20 mM Tris-HCl (pH 7.6), 0.01% sodium dodecyl sulfate, 30% formamide] containing either the universal probe EUB388 AlexaFluor 488 labeled probe or the negative control (nonsense) NONEUB388 AlexaFluor 546 (Thermo Fisher Scientific, BA, Arg). Stringent washing was performed by incubating the slide in washing buffer [20 mM Tris-HCl (pH 7.6), 0.01% sodium dodecyl sulfate, 112 mM NaCl] at 48°C for 15 min in a moist chamber a number of times, and subsequently rinsed with ddH2O for 5min. Finally, probed-hybridized tonsil cryosections were subjected to immunofluorescence staining as described above to determine the distribution of bacteria within the different tissue compartments. Images were acquired with a Nikon Eclipse Ti-E microscope (Nikon Instruments Inc., Tokyo, Japan) or an Olympus FV1000/IX81 confocal microscope (Olympus Corporation, Tokyo, Japan) using an oil-immersion objective (60X; numerical aperture, 1.42).

It is long established that B-cell–derived TNFα plays a crucial role in the development of follicular dendritic cells (FDCs) and B-cell follicles in spleen (9–11). TNFα and IL10 are cytokines with antagonistic actions. Interestingly, tonsils from OSA patients present a defective Breg compartment which correlates with higher proportion of GC B lymphocytes [BGC (6)], larger GC (12), and increased numbers of T follicular helper cells (Tfh) (13) than tonsils excised by other pathologies. Taking all into account, it would be expected that OSA tonsils exhibit a significant lymphocyte–derived TNFα compartment. We examined TNFα expression at the single cell level by FACS, upon TMC culture. A set of cultures was treated for 24 hs with IL2 and IL4 (IL2+IL4), which promote survival of all lymphocyte subsets through slight stimulation (14). Another set of cultures was supplemented for 24 hs and 48 hs with CpG and CD40L (in addition to IL2 and IL4, IL2+IL4+CD40L+CpG) to target stimulation towards B cells, aiming to assess specifically their capacity to contribute to the tonsillar pool of TNFα. Single cells were selected based on FSC-A vs FSC-H analysis (Singlets gate, Figure 1A ). The cells that were already dead prior to permeabilization were excluded from the analysis through a fixable dead cell staining (Viable gate, Figure 1A and Supplementary Figure 1 ). In agreement with previous observations (14), elevated levels of cell death were linked to stimulation of tonsillar cultures ( Figures 1A, B ), dominated by a variety of terminally differentiated and highly activated B cells. In fact, we used this decay in the viability of the TMC cultures to monitor for suitable activation. In our experience, TMC culture stimulation is quite susceptible to minimal changes in experimental conditions (cellular density or FCS and activation cocktail batches, for instance). Therefore, when comparing cytokine secretion within a cohort of patients, those cultures that did not present such evolution in terms of viability, proportion of CD3⁺ and CD20⁺ populations and CD20 down-modulation ( Figure 1A ), were not considered. Also, CD3⁺ cells served as an internal control for cytokine expression ( Figure 1C ). As expected, we observed that TMC from hypertrophied tonsils produced considerable amounts of TNFα when stimulated. At 24 hs post stimulation, approximately one third (29% ± SD 22%) of the cells from the IL2+IL4 stimulated cultures, expressed TNFα and that percentage reached ~40% (37%± SD 16%) in CD40L+CpG+IL2+IL4 stimulated cultures. The latter cultures were particularly affected by the surface CD20 down-modulation which takes place in response to general stimuli (15), albeit more pronounced with CD40L stimulation, as it has been extensively described previously (6, 16). Interestingly, B cells (CD20⁺) expressing TNFα were those presenting lower levels of CD20 (CD20^down TNFα⁺ cells, Figure 1A ) suggesting that TNFα expression might be another functional consequence of CD20 modulation and the downstream signaling pathways triggered post internalization. These findings confirmed that the extent of CD20 down-regulation correlates with the degree of B cell activation ( Figures 1A, C ) (6, 15, 16). Despite the fact that lymphoid populations other than CD20⁺ and CD3⁺ cells in TMC are negligible ( Figure 1A , non-cultured panel, double negative cells), their putative expansion, under the culture conditions used, needed to be tested as they would represent a putative contaminant of the CD20^down cells. Tonsillar dendritic cells (tDC) identified as CD20^-CD3^-CD11c⁺ appeared to increase their percentage within the CD20⁺CD20^down gate under stimuli. Even so, they represented 0.4%-0.7% of the latter, depending on the stimulus ( Supplementary Figure 2A ). Moreover, we could not detect any CD11b⁺ cells among the cultured TMC ( Supplementary Figure 2B ). The proportion of innate lymphoid cells (ILC) was also assayed ( Supplementary Figure 3 ) and proved to be negligible as well (detailed in the next section).

In order to compare TNFα expression by CD20^+/down population with that of CD3⁺ control cells within the same sample, we measured the median fluorescence intensity (MFI) for TNFα⁺ cells on both populations ( Supplementary Figure 4 ). Regardless of the type and length of stimulation, average MFI for CD3⁺ cells resulted around twice the same parameter for CD20^+/down cells ( Figure 1D ). We next determined in which proportion CD20^+/down and CD3⁺ cells were contributing to the total TNFα⁺ population ( Figure 1E ). Comprehensibly, at 24 hs post stimulation with CD40L+CpG+IL2+IL4, when B cell activation peaked, CD20^+/down cells represented the majority of tonsillar TNFα ⁺ cells (52.4% ± SD 13.4% CD20^+/down cells vs 41.7% ± SD 12.8% CD3⁺ cells). On the other hand, at 48 hs post stimulation with CD40L+CpG+IL2+IL4, when tonsillar B cells showed strongly reduced viability, tonsillar TNFα⁺ population was dominated by CD3⁺ cells (30.0% ± SD 11.7% CD20^+/down cells vs 63.0% ± SD 15.9% CD3⁺ cells). The cultures stimulated for 24 hs only with IL2+IL4 presented a tendency towards a slightly higher CD20^+/down cells contribution to the TNFα⁺ cell pool than CD3⁺ cells ( Figure 1F ). Importantly, the ability of tonsillar B cells to express TNFα was proved to be independent on the presence of the T cells in the TMC sample ( Supplementary Figure 5 ). Finally, we also tried stimulating the TMC cultures using as a surrogate antigen F(ab′)2 goat antibodies specific for IgM and IgG (in addition to CD40L+IL2+IL4) (14), obtaining very similar results to those of CD40L+CpG+IL2+IL4 stimulation (data not shown). Kim et al. (17) have previously reported that upon culture, TMC from hypertrophied tonsils rendered higher levels of TNFα⁺ in the supernatant than TMC from children with recurrent tonsillitis. These results extend such findings, indicating that being exposed to various stimuli, B cell population can largely contribute to the TNFα pool, supporting (if not promoting) tonsillar inflammation.

It has been suggested that development of OSA is associated with peripheral Th17/Treg imbalance and is characterized by a pro-inflammatory cytokine microenvironment (2). We have previously reported on the ability of tonsillar B cells to secrete IL17 (6). However, while B cells as a population can dominate tonsillar TNFα production, we found IL17 secretion is indeed controlled by CD3⁺ cells at every culture condition tested, including those specifically stimulating B lymphocytes ( Figures 2A, B ). TMC cultures supplemented with IL2+IL4 as well as those supplemented with CD40L+CpG+IL2+IL4 rendered an IL17⁺ population which comprised ~ 90% CD3⁺CD4⁺IL17⁺ cells at 24 and 48 hs. We confirmed this observation by measuring cytokines on the supernatant of some of the cultures by multiplex biomarker immunoassay (Luminex). We tried 4 different conditions to specifically stimulate B cells (CpG; CD40L; anti-(IgM/IgG)+CpG; anti-(IgM/IgG)+ CD40L), all of them in addition to IL2+IL4. We also added a culture treated only with IL2+IL4, as control. IL17 concentration did not significantly varied with culture condition compared to the control, as opposed to that of TNFα, which exhibited significant increments under B cell stimulating conditions (data not shown). As expected, CD3⁺IL17⁺ cells were confirmed to be CD3⁺CD4⁺IL17⁺ (Th17) cells ( Figure 2B ). Moreover, we tried to estimate the putative contribution of cells other than Th17 and B lymphocytes, to the IL17⁺ cell population pool. ILC3 are characterized by their capacity of secreting IL17 when stimulated. ILC3 represented around 0.03% of the TMC lymphoid gate (P3, Supplementary Figure 3 ). Their negligible fraction precluded us from detecting whether the culture conditions used stimulated them to express IL17. The same applied to Natural Killer (NK) cells which collectively signified 0.2% of the TMC lymphoid gate (P3, Supplementary Figure 3 ) but only a sub-fraction would be able to express IL17.

Th17 cells have appeared to exhibit plasticity of function and often co-produce other pro-inflammatory cytokines, particularly at sites of inflammation. Hence, we investigated whether tonsillar Th17 also expressed other cytokines as TNFα and interferon γ (IFNγ). The pseudo-color dot plots shown in Figure 2C were gated on CD3⁺ cells and show CD3 vs IL17 (upper panel) and IL17 vs TNFα (lower panel) staining. We detected co-expression of TNFα in a significant fraction of the Th17 cell population upon all culture conditions tested. At 24 hs post stimulation, nearly half of the Th17 population (44% ± SD 14%) from the IL2+IL4 stimulated cultures co-expressed TNFα. This percentage significantly increased when TMC were cultured with CD40L+CpG+IL2+IL4 for 24 hs (59% ± SD 14,5%) as well as for 48 hs (60%± SD 18%) indicating a positive correlation between higher frequencies of TNFα and IL17 and TNFα co-expression ( Figure 2D ). Low levels of expression of IFNγ observed upon the stimulating conditions tested precluded us from definitive conclusions regarding co-expression of IFNγ and IL17.

We concluded that tonsillar IL17 was produced primarily by Th17, which largely coproduced TNFα. Interestingly, several studies have linked the Th17 pathway with formation of GCs in mice spleens and ectopic B cell follicles at sites of inflammation (18–20). On the whole, TMC from hypertrophied tonsils promptly exhibited a pro-inflammatory cytokine profile in culture.

Like all mucosae, tonsillar immune actions (anti- and pro- inflammatory) must be tightly regulated, to balance the protection against virulent germs and the tolerance to harmless flora and innocuous Ags entering with air and food. Collectively, the results described above and others we have previously published (6, 14) are in agreement with the notion that tonsillar hypertrophy is a result of an imbalance between regulatory and effector immune functions which in turn lead to a local chronic inflammation. To this point, we have examined isolated cells. However, in vivo, immune processes occur in a complex tissue context which has to be comprehended. In order to recognize the histologic compartments within tonsils, we performed immune-fluorescence staining on cryosections from excised tonsils. The palatine tonsils are secondary lymphoid organs belonging to the mucosal associated lymphoid tissue (MALT), with unique histological characteristics. Unlike the lymph nodes, tonsils do not have afferent lymphatics, but present deep and branched crypts. Ag sample occurs through their distinctive tissue architecture and this is why tonsils are called lympho-epithelial organs. The tonsillar epithelium provides protection as well as serving to transport foreign material from the lumen to the lymphoid compartment. To evidence this mucosal surface, we tested the tonsillar epithelial cells’ reactivity to antibodies against CD1d, since it has been reported its expression in other mucosal epithelial cells, like intestinal (21), epidermal keratinocytes (22) and vaginal epithelium (23). We confirmed CD1d expression in tonsillar epithelium. Anatomically, the surface epithelium of the palatine tonsils is an extension of the stratified squamous epithelium of the oral mucosa ( Figure 3A , Supplementary Figures 6, 7 ). Importantly, in the crypts, the stratified epithelium becomes narrower and laxly textured [reticulated epithelium as termed by Oláh in 1978 (24)] and highly infiltrated with lymphocytes ( Figures 3B–D ). We observed CD1d expression in reticulated epithelium as well as in stratified squamous epithelium. Nevertheless, the level of that expression did not appear uniform at all levels of the multilayered squamous epithelium. The basal cells, which maintained an orientation perpendicular to the basement membrane, presented negligible immunoreactivity to anti CD1d. Conversely, CD1d was expressed strongly in the epithelial cells of the supra-basal and intermediate layers, significantly declining its expression in the apical layers. A similar pattern of CD1d staining is observed in the vaginal stratified epithelium (23). Expression was restrained mostly to the cell membrane in all cases ( Figure 3 ). This epithelium was supported by a layer of connective tissue ( Figure 3A ) that separates it from the lymphoid component, and resulted negative for CD1d, with the exception of sporadic infiltrating lymphoid cells. Figures 3B, C show a region of crypts lined by reticulated epithelium. In fact, the crypt in Figure 3C , is lined by reticulated epithelium on one side and stratified squamous epithelium on the other. Two characteristics of the former are clear from the figures, and these are the desquamation of the upper cell layers and the absence or disruption of the basal layer of the epithelium. In this case, there is no boundary between the epithelium and the underlying lymphoid tissue. The cells in the intermediate layer were distorted and often separated from one another by the invading lymphoid cells, yet again CD1d exhibited high expression in this layer. The reticular epithelium trailed the curving shapes of the underlying follicles. To our knowledge, this is the first report on CD1d expression by tonsillar epithelium.

To visualize the lymphoid compartment, we stained for (IgM/IgG)⁺ B cells ( Figure 3 ). Our results confirmed previous reports (25). (IgM/IgG)⁺ B cells (and plasma cells, distinguish by their size) appeared located not only in follicles, GC and their mantle zone, but also uniformly scattered throughout the tonsillar tissues ( Figures 3A–F ), generally associated in clusters when infiltrating reticular epithelial tissue ( Figure 3D ). IgA⁺ B/plasma cells on the other hand, were mainly found outside the germinal centers, especially in the epithelium of crypts and rarely appear in the GCs ( Figure 3F ). Also, it can be observed the protrusion of lymphocytes and plasma cells from the sites of desquamation to the lumen ( Figures 3C, D ). IgG and IgM can be observed in its soluble form, accumulated on the apical surface of the epithelial layer ( Figures 3A, D ). It has been previously shown that nasopharyngeal epithelium widely expresses FcRn, which mediates IgG transport across the epithelial cell layer (26). On the other hand, it does not express pIgR, hence the transport of dimeric IgA is unfeasible through this kind of epithelia (27), making IgG the predominant protective isotype in tonsils. The histological findings we described above are in agreement with those reports. Moreover, FcRn and CD1d belong to the same family of proteins, the non-classical MHC class I molecules and both have been shown to display a restriction in their expression to specific tissues, namely epithelial cells from other mucosal sites (28). Finally, follicular hyperplasia was present in all samples as well as intensively active germinal centers ( Figure 4 ). Such hyperplasia is attributable to proliferation of BGC, as we and others have previously shown (6, 12, 29). In light of the results shown in previous sections, such misguided B cell response can potentially fuel the local inflammatory microenvironment, extending the impairment of immune homeostatic control.

The most evident potential source of a local persistent immune response would be a local persistent infection. However, as opposed to recurrent tonsillitis, tonsillar hypertrophy has been long considered of non-infectious etiology. Actually, hypertrophied tonsils serve as non-infected control for tonsils affected by recurrent tonsillitis in a number of publications, including relevant and recent ones in which such control should be crucial (30). It has been generally considered that bacterial cultures from hypertrophied tonsils largely reflect oropharyngeal colonization. It is actually quite complex to distinguish infection versus colonization, provided that tonsils, being part of the mucosal immune system, are constantly exposed to the environment.

In order to try to discriminate between commensals and pathogens we started by naming the bacterial species that we could retrieve alive from the core of the samples upon cauterization of the surface, trying to identify abundant, relentless and viable populations from the clinical tissue. We cultured the core tonsillar tissue of 44 children undergoing tonsillectomy due to OSA, whose ages ranged from 2 to 15 years old ( Table 1 ). Within our cohort of patients, the frequency of tonsillectomy varied with the age, peaking at 5-6 years old ( Figure 5A ). All samples rendered viable bacterial cultures except for one. At phylum level, the tonsil cultures were dominated by Firmicutes (predominantly from the genera Streptococcus and Staphylococcus in that order), Proteobacteria (mostly from the genera Neisseria and Haemophilus), Bacteroidetes (principally genera Prevotella), Actinobacteria and Fusobacterium ( Tables 2 and 3 ) in agreement with recent data obtained by next generation sequence (NGS) (31).

Considering that certain microorganisms like Streptococcus pyogenes (S. pyogenes), Streptococcus pneumoniae (S. pneumoniae), Moraxella catarrhalis (M. catarrhalis), Haemophilus influenzae (H. influenzae) and Staphylococcus aureus (S. aureus) had been found either causing ear, nose and tonsil (ENT) pathology or as harmless local flora, both situations in competent hosts, we wanted to investigate their prevalence in our patients. We did not detect M. catarrhalis within our cohort of patients. We isolated S. pneumoniae from two patients. On the other hand, S. aureus, isolated from 45% of patients, accounted for the most frequent potentially pathogenic bacteria among the tonsils we analyzed, closely followed by H. influenzae (present in 27% of the patients) and S. pyogenes (on 23% of the samples). Bacterial isolates potentially pathogenic from the excised samples are shown in Figure 5B . In Argentina, vaccination against H. influenzae type b, S. pneumoniae, Neisseria meningitidis (N. meningitidis) and Corynebacterium diphtheria, which have been long known as commensal species in tonsils, is mandatory. We did not detect any patient carrying Corynebacterium diphtheriae but we found Neisseria meningitidis in the tonsils of a single patient and we did not serotype any isolate of H. influenzae.

In most cases we obtained co-isolates of different bacteria. Only 4 patients rendered cultures of single species, 3 of those were Streptococcus pyogenes group A. Co-isolates ranged from 2 to 7 species. Details of the mean of species co-isolated from each child and all isolated organisms from the samples are listed in Tables 1 – 3 . As we isolated a plethora of viable bacteria considered commensal and potential pathogenic as well, we were still unable to discriminate whether the tonsillar hypertrophy is of infectious nature.

The basic distinction between a pathogen and a commensal is that the first one displays aggressive tools for invasion. Usually this represents that such pathogen is able to penetrate the epithelial barriers. Therefore, we set out to detect bacterial presence in tonsillar biopsies by FISH, which allowed us to investigate its distribution and organization in situ. We used a fluorescent universal eubacterial (EUB338) probe followed by immune-fluorescence staining on cryosections from excised tonsils. As shown in Figures 5C, D , we detected bacterial aggregates not only associated with the surface of the epithelium but also within the lymphoid compartment, having breached the reticular epithelium. Therefore, while we cannot ascertain that the microorganisms detected in situ as well as through culture are the initiators of the ongoing inflammatory response, we did evidence that the chronification of the process must be certainly related to bacterial spreading beyond the normal boundaries. Also, such invasion is not associated to a single species as it can be observed in different patients harboring different colonizers, which have become pathogens, given the host´ requirement of the surgery to restore homeostasis.