You are here

Article Text

Article menu
Inflammatory bowel disease
Serum amyloid A for predicting prognosis in patients with newly diagnosed Crohn’s disease
  1. Qia Chen1,2,
  2. Xi Zhang1,
  3. Yizhe Tie1,
  4. Jianwu Zhang2,
  5. Pinwei Huang2,
  6. Yuxuan Xie2,
  7. Liqian Zhang1,
  8. Xueer Tang2,
  9. Zhirong Zeng1,
  10. Li Li1,
  11. Minhu Chen1,
  12. Rirong Chen1,
  13. Shenghong Zhang1,3
  1. 1Gastroenterology, The First Affiliated Hospital of Sun Yat-sen University, Guangzhou, Guangdong, China
  2. 2Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, People's Republic of China
  3. 3Guangxi Hospital Division of The First Affiliated Hospital, Sun Yat-sen University, Nanning, People's Republic of China
  1. Correspondence to Dr Rirong Chen; chenrr29{at}mail.sysu.edu.cn; Dr Shenghong Zhang; zhshh3{at}mail.sysu.edu.cn

Abstract

Objective Serum amyloid A (SAA) was found to be positively correlated with the activity of Crohn’s disease (CD); however, its prognostic value remains uncertain. Here, we examined its predictive ability in newly diagnosed CD and explored genetic association.

Methods This retrospective cohort study included patients newly diagnosed as CD at the First Affiliated Hospital of Sun Yat-sen University between June 2010 and March 2022. We employed receiver operating characteristic curve, Cox proportional hazard regression models and restricted cubic splines to investigate the prognostic performance of SAA for surgery and disease progression. To assess possible causality, a two-sample Mendelian randomisation (MR) of published genome-wide association study data was conducted.

Results During 2187.6 person-years (median age, 28 years, 72.4% male), 87 surgery and 153 disease progression events were documented. A 100-unit increment in SAA level generated 14% higher risk for surgery (adjusted HR (95% CI): 1.14 (1.05–1.23), p=0.001) and 12% for disease progression (1.12 (1.05–1.19), p<0.001). Baseline SAA level ≥89.2 mg/L led to significantly elevated risks for surgery (2.08 (1.31–3.28), p=0.002) and disease progression (1.72 (1.22–2.41), p=0.002). Such associations were assessed as linear. Adding SAA into a scheduled model significantly improved its predictive performances for surgery and disease progression (p for net reclassification indexes and integrated discrimination indexes <0.001). Unfortunately, no genetic causality between SAA and CD was observed in MR analysis. Sensitivity analyses showed robust results.

Conclusion Although causality was not found, baseline SAA level was an independent predictor of surgery and disease progression in newly diagnosed CD, and had additive benefit to existing prediction models.

  • CROHN'S COLITIS
  • GASTROINTESTINAL SURGERY
  • INFLAMMATION

Data availability statement

Data are available upon reasonable request. The datasets generated for this study are available on request from the corresponding authors.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

https://doi.org/10.1136/bmjgast-2024-001497

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    WHAT IS ALREADY KNOWN ON THIS TOPIC

    • Serum amyloid A (SAA) is an acute phase protein that can regulate immunity and lipid metabolism.

    • SAA has been found to be associated with clinical and endoscopic activity of Crohn’s disease (CD).

    • Prognostic value and causality remained to be investigated.

    WHAT THIS STUDY ADDS

    • Higher baseline SAA level was significantly associated with bowel resection, the risk increased linearly with increasing SAA.

    • In those with haemoglobin <120 g/L, such association was weakened.

    • For disease progression, a similar pattern was observed.

    • This prognostic association was merely correlational rather than causal.

    HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

    • SAA has the potential as a CD prognostic biomarker.

    • SAA could be a candidate predictor for future prediction models or artificial intelligence studies.

    • The mechanism causing poor CD prognosis remains to be elucidated.

    Introduction

    Crohn’s disease (CD), a type of inflammatory bowel disease (IBD), is featured with a repeating relapse-remission course and highly heterogenous prognosis, ranging from quiescent disease to complicated course with stricturing and/or penetrating lesions requiring abdominal surgery.1 Prognosis prediction is crucial for personalised patient management in CD, including selecting appropriate treatment regimens, determining assessment methods for disease activity and developing long-term follow-up plans. The currently known risk factors are insufficient to offer comprehensive prognostic information in patients with CD. Thus, exploring novel and reliable prognostic factors is essential. Biomarkers offer the advantages of being non-invasive, reproducible, cost-effective and providing a good reflection of the pathophysiological state.2 Identifying biomarkers that predict CD prognosis can enable the implementation of more personalised management strategies.

    Serum amyloid A (SAA) is an acute phase response protein regulating immunity and lipid transport, which is related to the pathogenesis of IBD.3 4 Overall, SAA exhibits a proinflammatory role in IBD due to its effects on immune cells.3 High SAA levels have been found to distinguish patients with elevated disease activity, notably even among those with normal C-reactive protein (CRP) levels.3 However, whether SAA could be a prognostic factor for CD remains unclear.

    Mendelian randomisation (MR) uses genetic variants as instrumental variables (IVs) to assess causal associations between exposure factors and outcomes.5 Its principal advantage lies in the fact that single nucleotide polymorphisms (SNPs) associated with exposure factors are assigned to offspring through genetic randomisation.5 This mechanism serves to mitigate the influence of confounders typically encountered in observational studies and enables the verification of reverse causality,6 making MR suitable for investigating causality between SAA and CD prognosis.

    In this study, we examined the prognostic value of SAA in predicting surgery and disease progression in patients with newly diagnosed CD. The added value of SAA to established risk factors for poor CD outcomes was also assessed. Moreover, we conducted a two-sample MR study to determine whether a causality exists.

    Methods

    Study design and population

    This retrospective cohort study included patients newly diagnosed as CD at the First Affiliated Hospital of Sun Yat-sen University between June 2010 and March 2022. Patients were excluded if they met any of the following criteria: (1) lack of SAA assessment within 1-month postadmission; (2) age <18 years at diagnosis; (3) missing baseline demographic and/or clinical data; and (4) acute infectious disease at admission.

    Data collection

    The SAA level at diagnosis was the primary variable of interest in this study. Serum samples were collected and tested within 1 month after admission. SAA levels were assessed continuously using the N latex SAA kit (Siemens Healthcare Diagnostics Products GmbH, Germany). Patients were classified using two distinct methods: (1) by SAA quartiles into low (<Q1), intermediate (Q1–Q3) and high (≥Q3) groups; (2) by the optimal cut-off from receiver operating characteristic curve analysis, corresponding to the maximum Youden’s index, into two separate groups.

    Other variables of interest included baseline demographic, clinical and biochemical data, detailed as follows: sex, age at diagnosis, body mass index (BMI), disease behaviour (inflammatory (B1), stricturing (B2) or penetrating (B3)), disease location (ileal (L1), colonic (L2) or ileocolonic (L3)), upper gastrointestinal involvement, perianal lesion, smoking history, surgical history, initial medication (biologics, corticosteroids, immunomodulators or other), and CRP, albumin, and haemoglobin levels.

    Outcomes

    The primary outcome was CD-related abdominal surgery, defined as any type of colectomy, ileocolectomy, small intestine resection or gastrectomy. The secondary outcome was disease progression, defined as occurrence of either newly developed stricturing or penetrating lesions (B1 to B2/B3 or B2 to B3), or CD-related abdominal surgery. The follow-up period began with CD diagnosis and ended with occurrence of outcomes or censoring. Censoring was defined as the absence of any outcomes at the last follow-up.

    Statistical analysis

    Categorical variables were summarised as frequencies with percentages while continuous variables as medians (Q1–Q3). Pearson’s χ2 test and Kruskal-Wallis test were employed to inspect differences among groups in categorical and continuous variables, respectively.

    The cumulative incidences of outcomes were estimated using Kaplan-Meier’s method, and differences among groups were compared using the log-rank test. For evaluating relationships between variables and outcomes, HRs/adjusted HRs (aHRs) and 95% CIs were computed by the Cox proportional hazards regression analysis. To account for probable confounders, two multivariable models were constructed: model 1 was adjusted for sex and age at diagnosis, whereas model 2 included further adjustments for disease behaviour, surgical history, smoking history and initial biological treatment. Linear trend test was performed using ternary SAA level as a continuous variable in the Cox regression model. Restricted cubic spline regression model with three knots was used to estimate the non-linear relationships between SAA and outcomes. Subgroup analyses were conducted to assess the interaction effects, stratified by the following characteristics: sex (male and female), age at diagnosis (<40 and ≥40 years), disease location (L1 and L2/L3), BMI (<18.5 and ≥18.5 kg/m2), initial biological treatment (yes and no) and haemoglobin level (<120 and ≥120 g/L).

    For assessing the additional benefit of incorporating SAA into established risk factors for CD prognosis prediction, we developed two prediction models: one was fitted based on age at diagnosis ≥40 years, disease behaviour, disease location and initial biological treatment; the other contained factors above and SAA. Thereafter, the added benefit of SAA for prediction was evaluated by Harrell’s concordance index (C-index), net reclassification index (NRI) and integrated discrimination index (IDI).

    Additionally, various sensitivity analyses were performed. First, we determined SAA cut-off values using several methods, including the median, quartiles, knot from the restricted cubic spline regression model and the cut-off value corresponding to the minimum p value in the log-rank test. Second, we further adjusted for perianal lesion in model 2. Third, we evaluated the prognostic value of SAA for disease modification, defined as newly developed stricturing or penetrating lesions (B1 to B2/B3 or B2 to B3).

    A bidirectional two-sample MR was conducted to explore the causal relationship between SAA levels and the prognosis of CD. The genetic data for SAA1 and SAA2 was extracted from a large-scale genome-wide association study (GWAS) involved 10 708 European participants.7 For poor prognosis of CD, which was defined as the necessity for sequential therapy with two or more immunomodulators, undergoing two or more abdominal operations, or a combination thereof, a publicly available GWAS involved 2734 European participants was selected for MR analysis.8 We selected IVs with p<5E-8 and excluded linkage disequilibrium (r²<0.001, with a 10 000 kb window size). Then we calculated the F statistics (F=beta²/SE²) for every SNP. F statistic <10 indicates the presence of weak IVs, which would be excluded from IVs in the analysis. Furthermore, we checked every SNP in PhenoScannerV2 (http://www.phenoscanner.medschl.cam.ac.uk/) to exclude potential confounders (sex, BMI and smoking).9 The primary analytical method was inverse variance weighted method, with additional sensitivity analyses conducted through MR-Egger, weighted median, weighted mode and simple mode methods to ensure robustness of our findings.10 Heterogeneity was evaluated by Cochran’s Q test and pleiotropy was assessed by MR-Egger regression.

    All analyses were conducted in R V.4.2.3 (R Foundation for Statistical Computing, Vienna, Austria). A two-sided p<0.05 was considered indicating statistical significance.

    Results

    Baseline characteristics

    Altogether 602 (male, n=436, 72.4%) of the 871 patients were included in this study (online supplemental figure 1). Table 1 presents the baseline characteristics. The median baseline SAA level was 62.60 (13.45–250.75) mg/L, and that of age at diagnosis was 28.00 (22.25–35.00) years.

    View this table:
    Table 1

    Baseline characteristics

    After ternary grouping, patients with lower SAA levels tended to be male (p=0.045), suffer from inflammatory and ileal diseases (p=0.034 and p<0.001, respectively), as well as be free from perianal lesions (p<0.001) than those with higher SAA levels. These patients also had significantly lower CRP levels and higher BMI, haemoglobin and albumin levels (all p<0.001), indicating milder inflammation and better nutrients (online supplemental table 1).

    Prognostic value of SAA for CD-related surgery and disease progression

    During a median follow-up of 37 months, CD-related abdominal surgery and disease progression occurred in 87 and 153 patients, respectively (online supplemental figure 2). A cut-off value of 89.2 mg/L for the SAA level allowed an optimal risk stratification for surgery, and the corresponding C-index (95% CI) was 0.621 (0.558–0.684) (online supplemental figure 3).

    In the survival analysis, contrasted with those with low SAA levels, patients with SAA levels ≥89.2 mg/L exhibited a higher cumulative incidence of surgery at each time point (1 year, 10.6% vs 4.1%; 3 years, 14.1% vs 5.7%; 5 years, 18.3% vs 6.9%; 10 years, 19.7% vs 9.4%; all p<0.001; figure 1; online supplemental table 2). Analogous phenomenon was observed for disease progression (online supplemental table 2).

    Figure 1
    Figure 1

    Cumulative incidences of surgery (A) and disease progression (B) stratified by binary SAA level. SAA, serum amyloid A.

    The fully adjusted Cox regression model uncovered that an SAA level ≥89.2 mg/L was related to a 108% higher risk for surgery (aHR (95% CI): 2.08 (1.31–3.28), p=0.002) and 72% for disease progression (1.72, (1.22–2.41), p=0.002; table 2) comparing with lower level. Similarly, each 100-unit increment in SAA level led to a 14% elevated risk for surgery (1.14 (1.05–1.23), p=0.001) and 12% for disease progression (1.12 (1.05–1.19), p<0.001). For ternary comparison, only the high SAA group exhibited significantly increased risks for surgery (2.82 (1.42–5.58), p=0.003) and disease progression (2.10 (1.30–3.37), p=0.002; table 2) compared with the low SAA group. In the restricted cubic spline analysis, SAA level demonstrated positive linear associations with surgery (p for non-linearity=0.231) and disease progression (p for non-linearity=0.362; figure 2), consistent with the results above.

    Figure 2
    Figure 2

    Linear relationship between baseline SAA level and surgery (A) and disease progression (B). The HRs were computed based on a median baseline SAA level of 61.1 mg/L. The p values for non-linearity were 0.231 (A) and 0.362 (B), respectively, indicating that there was no non-linear relationship between SAA and either outcome. SAA, serum amyloid A.

    View this table:
    Table 2

    Risks of surgery and disease progression stratified by SAA level

    In the subgroup analyses, SAA had no significant interaction with sex, age at diagnosis, disease location, BMI and initial biological treatment (p for interaction >0.05; online supplemental figure 4). However, the correlation between SAA level and risk for surgery was more prominent in patients with haemoglobin levels ≥120 mg/L (HR (95% CI): 3.72 (1.63–7.50), p=0.002) than in others (1.31 (0.77–2.22), p=0.326; p for interaction <0.001). A similar phenomenon was observed for disease progression.

    Added value of SAA to the model with known risk factors

    Table 3 shows the C-index, NRI and IDI of the original and SAA-added models. Incorporating SAA into the model with known risk factors significantly improved its prognostic ability. For surgery, the NRI (95% CI) and IDI (95% CI) severally were 0.193 (0.018–0.276, p<0.001) and 0.022 (0.001–0.069, p<0.001), respectively; while for disease progression, they were 0.065 (0.004–0.195, p<0.001) and 0.017 (0.002–0.034, p<0.001; table 3), respectively. These results confirmed that the addition of SAA significantly enhanced the predictive power of the original model.

    View this table:
    Table 3

    Performance of SAA and known risk factors in predicting surgery and disease progression

    Sensitivity analyses

    Akin to the primary analysis, higher SAA levels were still correlated to higher risks for occurring surgery and disease progression after replacing the cut-off value (online supplemental table 3). SAA kept stable ability to predict outcomes after further adjusting for perianal lesion in the model 2 (online supplemental table 4). When predicting disease modification, baseline SAA level ≥89.2 mg/L was associated with a 61% higher risk (aHR (95% CI): 1.61 (1.09–2.38), p=0.016; online supplemental table 5). Likewise, adding SAA to the scheduled model significantly improved its power to predict disease modification, with NRI (95% CI) and IDI (95% CI) of 0.062 (0.003–0.187, p<0.001) and 0.011 (0.001–0.041, p<0.001; online supplemental table 6).

    Causality between SAA and CD prognosis in MR

    In the GWAS cohorts (online supplemental table 7), after selecting for p<5E-8 and excluding the interference of linkage disequilibrium, four IVs for SAA1, five IVs for SAA2 and three IVs for CD prognosis have been obtained. No confounders were found during this process. Besides, the F statistics of IVs range from 30.0 to 1038.3, indicating no weak IVs in our analysis (online supplemental tables 8–10). All five analytical methods indicated no causality between the two members of SAA and poor prognosis in CD (online supplemental figures 5 and 6). No heterogeneity or pleiotropy has been observed during this process.

    Discussion

    Predicting CD prognosis has guiding significance for clinicians to better manage patients and make therapeutic decisions, and understanding its affecting factors is helpful to intervene and improve CD prognosis. SAA is related to clinical and endoscopic activity of CD, and has a potential as a prognostic biomarker.3 In this study, we discovered that the baseline SAA level in patients with newly diagnosed CD was significantly positively associated with risks for CD-related abdominal surgery and disease progression, and the optimal cut-off value for surgical risk stratification was 89.2 mg/L. Incorporating SAA into the model with established risk factors significantly improved its predictive power. Such prognostic values of SAA were verified as robust by sensitivity analyses. Although no causality between SAA and CD prognosis was found, these findings still indicated that SAA can serve as a biomarker for predicting CD prognosis and aiding clinical practice.

    Currently, the number of studies focusing on the application of SAA in CD is limited, but these studies overall achieved good results.3 11–13 In a cohort of 94 patients with CD, SAA demonstrated strong ability to assess endoscopic activity, and could still identify 70% of active patients even when CRP was negative.11 Patients with SAA >5.9 µg/dL were observed more possible to experience clinical relapse in another cohort of 41 CD cases.12 One additional study on 118 IBD patients showed that combining SAA with interleukin 6 (IL-6), IL-8 and eotaxin-1 could improve its recognition ability of endoscopic activity, achieving a C-index of 0.84.13 On their basis, we specifically focused on the prognostic value of SAA and uncovered that it exhibited positive linear relationships with both surgery and disease progression. In addition, a significant interaction between SAA and haemoglobin levels was observed; high SAA levels markedly indicated poor prognosis in patients with haemoglobin ≥120 g/L.

    Prior studies have identified various risk factors for CD prognosis.14–20 In a 200 CD cohort, Henriksen et al reported that stricturing or penetrating disease at diagnosis was a risk factor for surgery during a 5-year follow-up, while age <40 years was a protective factor.15 Another multi-centre study on 345 patients identified pure small bowel disease as a risk factor for surgery within 6 months, whereas colonic disease manifested an inverse association.17 Analysis of seven randomised controlled trials (five on CD and two on ulcerative colitis) determined that initial biological treatment was a protective factor for surgery.20 In the current study, we fitted a model based on the existing evidences, and found that SAA could further elevate the predictive performance of the original model. This finding identified SAA as a candidate predictor for future prediction models or artificial intelligence studies, thereby providing more comprehensive information for clinicians.

    To explore a possible causal relationship between SAA and poor prognosis in newly diagnosed CD, we conducted an MR analysis. MR has been extensively applied in investigating the aetiology of CD.21–23 However, its application in exploring prognostic factors related to CD is relatively limited. Through MR analysis, we found no evidence of causality between SAA and CD prognosis in the European cohort. Considering that SAA is a marker of acute inflammation, this suggested no direct causality between SAA and long-term prognosis of CD. Nevertheless, the significant correlations between SAA and outcomes still make it promising for future prognostic studies and clinical practices. Moreover, a limited number of SNPs were identified in our MR analysis; hence, further research with larger-scale GWAS could be considered to elucidate the relationship between SAA and CD prognosis.

    Previously, we reviewed the available evidence related to the mechanism of SAA affecting IBD.3 In brief, SAA has a dual role in IBD. On the one hand, SAA modulates the expression of IL-22 in intestinal neutrophils to protect the intestinal epithelium,24 enhances the phagocytosis of gram-negative bacteria by phagocytic cells25 26 and promotes tissue repair by mediating lipid transportation during inflammation.27–32 On the other hand, SAA not only activates the proinflammatory activity of macrophages,4 33 34 but also induces the differentiation of naive T cells into pathogenic T-helper 17 cells, thereby aggravating inflammation.35 In humans, as the SAA3 gene is a pseudogene, SAA should mainly exhibit proinflammatory effects.3 Positive linear associations between baseline SAA level and outcomes were demonstrated in our study, consistent with both our hypothesis and preceding studies.3 11–13

    Our study first clarified the prognostic function of SAA in a cohort of more than 100 CD patients, and our results have important clinical implications for personalised patient management in CD. First, the positive linear relationships between SAA and outcomes suggested that SAA could be an effective biomarker for risk stratification in newly diagnosed CD patients. In addition, SAA may further improve the predictive power of existing models for CD prognosis. Finally, the sensitivity analyses yielded robust results, confirming the prognostic value of SAA.

    However, several limitations are inevitable. First, we were not able to perform a power analysis because of the lack of previous studies, and although we considered all patients in the study period, it is unknown whether the sample size of this cohort was sufficient for adequate statistical power. Second, the single-centre retrospective design limits the universality of the results. Third, only adults with CD were included; hence, in paediatric CD patients, the prognostic value and optimal cut-off value of SAA necessitate further research. Fourth, the effect of possible confounders, such as alcohol consumption, sleeping patterns, genetic polymorphism, were not included in our study, although we have used multivariable Cox regression to adjusted confounders regrading demographic, clinical and treatment-associated characteristics. Finally, in the MR analysis, we found no published GWAS data from Asian population was available, and in European cohorts, only a few SNPs were identified, which might limit the search for causality.

    In conclusion, this study demonstrates that elevated baseline SAA level was an independent risk factor for surgery and disease progression in patients with newly diagnosed CD, and SAA might own significant additive benefit to existing prediction models. Although the association between SAA and CD prognosis was assessed to be correlational rather than causal, these findings supported that SAA could be a useful biomarker for CD prognosis, which can contribute to personalised patient management and clinical decision-making. More studies could be considered to either determine a widely applicable cut-off value for risk stratification or explore the possible causality thoroughly. Furthermore, the application of SAA in ulcerative colitis is also worth exploring.

    Data availability statement

    Data are available upon reasonable request. The datasets generated for this study are available on request from the corresponding authors.

    Ethics statements

    Patient consent for publication

    Ethics approval

    This study involves human participants and this study was approved by the Institutional Ethics Committee for Clinical Research and Animal Trials of the First Affiliated Hospital of Sun Yat-sen University (No. IIT-2023-247). All patients who participated in this study provided written informed consent at admission for use of their data for research purposes.

    References

    1. ↵
      1. Roda G,
      2. Chien Ng S,
      3. Kotze PG, et al
      . Crohn’s disease. Nat Rev Dis Primers 2020;6:22. doi:10.1038/s41572-020-0156-2
      OpenUrl
    2. ↵
      1. Goodsall TM,
      2. Noy R,
      3. Nguyen TM, et al
      . Systematic Review: Patient Perceptions of Monitoring Tools in Inflammatory Bowel Disease. J Can Assoc Gastroenterol 2021;4:e31–41. doi:10.1093/jcag/gwaa001
      OpenUrlPubMed
    3. ↵
      1. Chen R,
      2. Chen Q,
      3. Zheng J, et al
      . Serum amyloid protein A in inflammatory bowel disease: from bench to bedside. Cell Death Discov 2023;9:154. doi:10.1038/s41420-023-01455-5
    4. ↵
      1. Sack GH
      . Serum amyloid A – a review. Mol Med 2018;24:46. doi:10.1186/s10020-018-0047-0
      OpenUrlCrossRefPubMed
    5. ↵
      1. Sekula P,
      2. Del Greco M F,
      3. Pattaro C, et al
      . Mendelian Randomization as an Approach to Assess Causality Using Observational Data. J Am Soc Nephrol 2016;27:3253–65. doi:10.1681/ASN.2016010098
      OpenUrlAbstract/FREE Full Text
    6. ↵
      1. Larsson SC,
      2. Butterworth AS,
      3. Burgess S
      . Mendelian randomization for cardiovascular diseases: principles and applications. Eur Heart J 2023;44:4913–24. doi:10.1093/eurheartj/ehad736
      OpenUrlCrossRef
    7. ↵
      1. Pietzner M,
      2. Wheeler E,
      3. Carrasco-Zanini J, et al
      . Genetic architecture of host proteins involved in SARS-CoV-2 infection. Nat Commun 2020;11:6397. doi:10.1038/s41467-020-19996-z
    8. ↵
      1. Lee JC,
      2. Biasci D,
      3. Roberts R, et al
      . Genome-wide association study identifies distinct genetic contributions to prognosis and susceptibility in Crohn’s disease. Nat Genet 2017;49:262–8. doi:10.1038/ng.3755
      OpenUrlCrossRefPubMed
    9. ↵
      1. Burgess S,
      2. Small DS,
      3. Thompson SG
      . A review of instrumental variable estimators for Mendelian randomization. Stat Methods Med Res 2017;26:2333–55. doi:10.1177/0962280215597579
      OpenUrlCrossRefPubMed
    10. ↵
      1. Lee YH
      . Overview of Mendelian Randomization Analysis. J Rheum Dis 2020;27:241–6. doi:10.4078/jrd.2020.27.4.241
      OpenUrl
    11. ↵
      1. Yarur AJ,
      2. Quintero MA,
      3. Jain A, et al
      . Serum Amyloid A as a Surrogate Marker for Mucosal and Histologic Inflammation in Patients with Crohn’s Disease. Inflamm Bowel Dis 2017;23:158–64. doi:10.1097/MIB.0000000000000991
      OpenUrl
    12. ↵
      1. Ishihara S,
      2. Tada Y,
      3. Kawashima K, et al
      . Serum amyloid A level correlated with endoscopic findings in patients with Crohn’s disease-Possible biomarker for evaluating mucosal healing. Dig Liver Dis 2018;50:553–8. doi:10.1016/j.dld.2017.12.010
      OpenUrl
    13. ↵
      1. Bourgonje AR,
      2. von Martels JZH,
      3. Gabriëls RY, et al
      . A Combined Set of Four Serum Inflammatory Biomarkers Reliably Predicts Endoscopic Disease Activity in Inflammatory Bowel Disease. Front Med (Lausanne) 2019;6:251. doi:10.3389/fmed.2019.00251
    14. ↵
      1. Solberg IC,
      2. Vatn MH,
      3. Høie O, et al
      . Clinical course in Crohn’s disease: results of a Norwegian population-based ten-year follow-up study. Clin Gastroenterol Hepatol 2007;5:1430–8. doi:10.1016/j.cgh.2007.09.002
      OpenUrlCrossRefPubMedWeb of Science
    15. ↵
      1. Henriksen M,
      2. Jahnsen J,
      3. Lygren I, et al
      . C-reactive protein: a predictive factor and marker of inflammation in inflammatory bowel disease. Results from a prospective population-based study. Gut 2008;57:1518–23. doi:10.1136/gut.2007.146357
      OpenUrlAbstract/FREE Full Text
    16. ↵
      1. Peyrin-Biroulet L,
      2. Harmsen SW,
      3. Tremaine WJ, et al
      . Surgery in a Population-Based Cohort of Crohn’s Disease From Olmsted County, Minnesota (1970–2004). Am J Gastroenterol 2012;107:1693–701. doi:10.1038/ajg.2012.298
      OpenUrlCrossRefPubMed
    17. ↵
      1. Sands BE,
      2. Arsenault JE,
      3. Rosen MJ, et al
      . Risk of early surgery for Crohn’s disease: implications for early treatment strategies. Am J Gastroenterol 2003;98:2712–8. doi:10.1111/j.1572-0241.2003.08674.x
      OpenUrlCrossRefPubMedWeb of Science
    18. ↵
      1. Ryan JD,
      2. Silverberg MS,
      3. Xu W, et al
      . Predicting complicated Crohn’s disease and surgery: phenotypes, genetics, serology and psychological characteristics of a population-based cohort. Aliment Pharmacol Ther 2013;38:274–83. doi:10.1111/apt.12368
      OpenUrlCrossRefPubMed
    19. ↵
      1. Singh S,
      2. Murad MH,
      3. Fumery M, et al
      . Comparative efficacy and safety of biologic therapies for moderate-to-severe Crohn’s disease: a systematic review and network meta-analysis. Lancet Gastroenterol Hepatol 2021;6:1002–14. doi:10.1016/S2468-1253(21)00312-5
      OpenUrl
    20. ↵
      1. Mao EJ,
      2. Hazlewood GS,
      3. Kaplan GG, et al
      . Systematic review with meta-analysis: comparative efficacy of immunosuppressants and biologics for reducing hospitalisation and surgery in Crohn’s disease and ulcerative colitis. Aliment Pharmacol Ther 2017;45:3–13. doi:10.1111/apt.13847
      OpenUrlPubMed
    21. ↵
      1. Saadh MJ,
      2. Pal RS,
      3. Arias-Gonzáles JL, et al
      . A Mendelian Randomization Analysis Investigates Causal Associations between Inflammatory Bowel Diseases and Variable Risk Factors. Nutrients 2023;15:1202. doi:10.3390/nu15051202
    22. ↵
      1. Parisinos CA,
      2. Serghiou S,
      3. Katsoulis M, et al
      . Variation in Interleukin 6 Receptor Gene Associates With Risk of Crohn’s Disease and Ulcerative Colitis. Gastroenterology 2018;155:303–6. doi:10.1053/j.gastro.2018.05.022
      OpenUrlCrossRef
    23. ↵
      1. Gao H,
      2. Zheng S,
      3. Yuan X, et al
      . Causal association between inflammatory bowel disease and 32 site-specific extracolonic cancers: a Mendelian randomization study. BMC Med 2023;21:389. doi:10.1186/s12916-023-03096-y
    24. ↵
      1. Zhang G,
      2. Liu J,
      3. Wu L, et al
      . Elevated Expression of Serum Amyloid A 3 Protects Colon Epithelium Against Acute Injury Through TLR2-Dependent Induction of Neutrophil IL-22 Expression in a Mouse Model of Colitis. Front Immunol 2018;9:1503. doi:10.3389/fimmu.2018.01503
    25. ↵
      1. Hari-Dass R,
      2. Shah C,
      3. Meyer DJ, et al
      . Serum amyloid A protein binds to outer membrane protein A of gram-negative bacteria. J Biol Chem 2005;280:18562–7. doi:10.1074/jbc.M500490200
      OpenUrlAbstract/FREE Full Text
    26. ↵
      1. Shah C,
      2. Hari-Dass R,
      3. Raynes JG
      . Serum amyloid A is an innate immune opsonin for Gram-negative bacteria. Blood 2006;108:1751–7. doi:10.1182/blood-2005-11-011932
      OpenUrlAbstract/FREE Full Text
    27. ↵
      1. van der Westhuyzen DR,
      2. Cai L,
      3. de Beer MC, et al
      . Serum amyloid A promotes cholesterol efflux mediated by scavenger receptor B-I. J Biol Chem 2005;280:35890–5. doi:10.1074/jbc.M505685200
      OpenUrlAbstract/FREE Full Text
    28. ↵
      1. Takase H,
      2. Furuchi H,
      3. Tanaka M, et al
      . Characterization of reconstituted high-density lipoprotein particles formed by lipid interactions with human serum amyloid A. Biochim et Biophys Acta (BBA) - Mol Cell Biol Lipids 2014;1841:1467–74. doi:10.1016/j.bbalip.2014.07.012
      OpenUrl
    29. ↵
      1. Jayaraman S,
      2. Haupt C,
      3. Gursky O
      . Paradoxical effects of SAA on lipoprotein oxidation suggest a new antioxidant function for SAA. J Lipid Res 2016;57:2138–49. doi:10.1194/jlr.M071191
      OpenUrlAbstract/FREE Full Text
    30. ↵
      1. Frame NM,
      2. Jayaraman S,
      3. Gantz DL, et al
      . Serum amyloid A self-assembles with phospholipids to form stable protein-rich nanoparticles with a distinct structure: A hypothetical function of SAA as a “molecular mop” in immune response. J Struct Biol 2017;200:293–302. doi:10.1016/j.jsb.2017.06.007
      OpenUrlCrossRefPubMed
    31. ↵
      1. Jayaraman S,
      2. Gantz DL,
      3. Haupt C, et al
      . Serum amyloid A sequesters diverse phospholipids and their hydrolytic products, hampering fibril formation and proteolysis in a lipid-dependent manner. Chem Commun (Camb) 2018;54:3532–5. doi:10.1039/C8CC01424H
      OpenUrl
    32. ↵
      1. Takase H,
      2. Tanaka M,
      3. Nakamura Y, et al
      . Effects of lipid composition on the structural properties of human serum amyloid A in reconstituted high-density lipoprotein particles. Chem Phys Lipids 2019;221:8–14. doi:10.1016/j.chemphyslip.2019.03.001
      OpenUrl
    33. ↵
      1. Ather JL,
      2. Ckless K,
      3. Martin R, et al
      . Serum amyloid A activates the NLRP3 inflammasome and promotes Th17 allergic asthma in mice. J Immunol 2011;187:64–73. doi:10.4049/jimmunol.1100500
      OpenUrlAbstract/FREE Full Text
    34. ↵
      1. Joosten LAB,
      2. Netea MG,
      3. Dinarello CA
      . Interleukin-1β in innate inflammation, autophagy and immunity. Semin Immunol 2013;25:416–24. doi:10.1016/j.smim.2013.10.018
      OpenUrlCrossRefPubMed
    35. ↵
      1. Lee J-Y,
      2. Hall JA,
      3. Kroehling L, et al
      . Serum Amyloid A Proteins Induce Pathogenic Th17 Cells and Promote Inflammatory Disease. Cell 2020;180:79–91. doi:10.1016/j.cell.2019.11.026
      OpenUrlCrossRefPubMed

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • QC, XZ, YT and JZ are joint first authors.

    • QC, XZ, YT and JZ contributed equally.

    • Contributors Guarantor of article: SZ. SZ, LL and MC: conceptualisation; funding acquisition; writing-review and editing. RC: conceptualisation; methodology; writing-original draft preparation; writing-review and editing; project administration. QC and YT: data curation; formal analysis; methodology; writing-original draft preparation; project administration. XZ and JZ: supervision; methodology; writing-original draft preparation; PH, YX, LZ and XT: project administration; data curation; writing-review and editing. ZZ: supervision; writing-review and editing. All authors approved the final version of the article.

    • Funding This work was supported by the National Natural Science Foundation of China (#82000520, #82270555, #82070538) and Guangdong Science and Technology Department (#2020A1515010249, #2021A1515220107). This study was funded by the China Crohn’s & Colitis Foundation (CCCF) under Grant No. CCCF-QF-2022B36-7.

    • Competing interests None declared.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.