Baseline TREM-1 Whole Blood Gene Expression Does Not Predict Response to Adalimumab Treatment in Patients with Ulcerative Colitis or Crohn’s Disease in the SERENE Studies

Abstract Background and Aims This study assessed whether baseline triggering receptor expressed on myeloid cells [TREM-1] whole blood gene expression predicts response to anti-tumour necrosis factor [anti-TNF] therapy in patients with ulcerative colitis [UC] or Crohn’s disease [CD]. Methods TREM-1 whole blood gene expression was analysed by RNA sequencing in patients with moderately to severely active UC or CD treated with adalimumab in the Phase 3 SERENE-UC and SERENE-CD clinical trials. The predictive value of baseline TREM-1 expression was evaluated and compared according to endoscopic and clinical response vs non-response, and remission vs non-remission, at Weeks 8 and 52 [SERENE-UC], and Weeks 12 and 56 [SERENE-CD]. Results TREM-1 expression was analysed in 95 and 106 patients with UC and CD, respectively, receiving standard-dose adalimumab induction treatment. In SERENE-UC, baseline TREM-1 expression was not predictive of endoscopic response [p = 0.48], endoscopic remission [p = 0.53], clinical response [p = 0.58], or clinical remission [p = 0.79] at Week 8, or clinical response [p = 0.60] at Week 52. However, an association was observed with endoscopic response [p = 0.01], endoscopic remission [p = 0.048], and clinical remission [p = 0.04997] at Week 52. For SERENE-CD, baseline TREM-1 expression was not predictive of endoscopic response [p = 0.56], endoscopic remission [p = 0.33], clinical response [p = 0.07], or clinical remission [p = 0.65] at Week 12, or endoscopic response [p = 0.61], endoscopic remission [p = 0.51], clinical response [p = 0.62], or clinical remission [p = 0.97] at Week 56. Conclusions Baseline TREM-1 gene expression did not uniformly predict adalimumab response in SERENE clinical trials. Further research is needed to identify potential blood-based biomarkers predictive of response to anti-TNF therapy in patients with inflammatory bowel disease. Clinicaltrials.gov identifiers NCT02065622; NCT02065570


Introduction
Ulcerative colitis [UC] and Crohn's disease [CD] are immunemediated inflammatory bowel diseases [IBDs] with a complex interplay of genetic and environmental factors involved in their pathogenesis. 1Adalimumab is an anti-tumour necrosis factor [anti-TNF] agent that is approved by the US Food and Drug Administration 2 and the European Medicines Agency 3 for the treatment of moderately to severely active UC and CD.Although anti-TNF agents such as adalimumab play an important role in the management of IBD, primary non-response to anti-TNFs occurs in 10-40% of patients. 4Identifying factors that predict response to anti-TNF therapy may improve the success of the treat-to-target approach recommended in IBD management guidelines. 5,6][9][10][11] In one study, baseline whole blood TREM-1 expression was downregulated in a mixed population of patients with UC or CD who achieved endoscopic remission following anti-TNF treatment, and low gut mucosa and whole blood expression of TREM-1 demonstrated similar accuracy as biomarkers of anti-TNF responsiveness; in contrast, a low serum TREM-1 protein level was not predictive of response. 7Another study showed that low baseline whole blood TREM-1 levels could be used as a potential predictive biomarker of anti-TNF therapeutic efficacy, in terms of mucosal healing in patients with CD. 8 Consistent with these results, further analysis found an association between high TREM-1 expression levels in cluster of differentiation 14-positive monocytes and decreased differentiation to M2-type regulatory macrophages, which are involved in mediating the anti-TNF response, indicating that response to anti-TNF agents may be linked to low TREM-1 levels in monocytes in patients with IBD. 9 In contrast to these findings, significantly downregulated baseline TREM-1 whole blood expression was predictive for non-response to anti-TNF therapy in a small cohort of patients with CD, based on clinical and/or endoscopic improvement of IBD-related symptom response criteria. 10Gene expression data from a cell-centred meta-analysis conducted by the authors of this latter study suggested that, in individuals who do not respond to anti-TNF agents, an increase in inflammatory macrophages is associated with upregulation of TREM-1 and chemokine receptor type 2-chemokine ligand 7 axes. 10Further, results from a study of paediatric patients with IBD suggested that baseline TREM-1 whole blood expression was not predictive for response to anti-TNF therapy, with no differences seen between responders and non-responders. 11One potential explanation for these opposing findings is the widespread use of corticosteroid [CS] drugs among patients with IBD. 12 The effect of CS use on TREM-1 expression is unclear and may be dependent on the individual CS.For example, in murine models, the CS dexamethasone has been shown to suppress TREM-1 expression on neutrophils, with this suppressive effect being mediated by TNF-alpha; in contrast, hydrocortisone had no effect on TREM-1 expression compared with controls. 13The differences in study designs, patient populations and their CS use, and findings mean that the potential utility of baseline TREM-1 whole blood gene expression as a predictive biomarker for response to anti-TNF therapy has yet to be elucidated.
The aim of the current study was to determine whether baseline TREM-1 whole blood gene expression was predictive of outcomes following standard adalimumab treatment in patients with UC or CD in the large, randomized, Phase 3 SERENE-UC and -CD studies. 14,15

SERENE-UC
[NCT02065622] 14 and SERENE-CD [NCT02065570] 15 were Phase 3, double-blind, randomized, multicentre clinical trials that evaluated higher vs standard adalimumab induction dosing regimens [Table 1] followed by adalimumab maintenance therapy in adult patients with moderately to severely active UC [n = 852] or CD [n = 514].The primary endpoint of SERENE-UC was clinical remission as per full Mayo score at Weeks 8 and 52.The co-primary endpoints of SERENE-CD were Clinical Disease Activity Index clinical remission at Week 4 and endoscopic response at Week 12.The full methods and results of SERENE-UC 14 and SERENE-CD 15 have been published previously.

Assessment of clinical outcomes
In the present study, clinical and endoscopic response and remission, defined in Table 2, were assessed at Weeks 8 and 52 in the SERENE-UC population, and at Weeks 12 and 56 in SERENE-CD participants.

Evaluation of TREM-1 expression
In patients who provided consent for exploratory biomarker analyses, whole blood TREM-1 gene expression was assessed by RNA sequencing [RNA-seq] at baseline and Weeks 2, 4, and 8 for UC, and at baseline and Weeks 2, 4, and 12 for CD.Single-end RNA-seq was performed to generate 76-bp reads.The samples were enriched by poly-A selection and globulin depletion with 14 Universal Human Reference RNA controls in 24 sequencing flow cells.Unique molecular identifier tagging was also used to improve sensitivity.Average unique read numbers were 20.6 million [standard error (SE): 5.04 million] in UC and 15.2 million [SE: 7.2 million] in CD.Soluble TREM-1 protein levels at baseline were also assessed using the Olink inflammation panel.

Statistical considerations
As this was a post hoc analysis, the sample sizes for each analysis were not pre-specified, and were instead determined by the numbers of patients from each trial who provided consent for exploratory molecular biomarker analyses.Unless otherwise specified, analyses were conducted only in the subset of patients who received standard-dose induction adalimumab.A univariate multiple regression analysis corrected for multiple testing with baseline TREM-1 gene expression [log 2 counts per million] as the response and baseline CS use as a predictor was performed using data from each trial separately to assess whether CS use impacted TREM-1 gene expression.
Baseline TREM-1 gene and soluble protein expression were analysed in patients evaluated for clinical outcomes at Weeks 8 and 52 in SERENE-UC, and at Weeks 12 and 56 in SERENE-CD.Two-sample t-tests were used to compare: • Baseline TREM-1 gene and soluble protein expression in clinical remitters vs clinical non-remitters, clinical responders vs clinical non-responders, endoscopic remitters   In addition, baseline TREM-1 gene expression was compared with a two-sample t-test in clinical and endoscopic remitters and responders vs non-remitters and non-responders stratified into quartiles based on Week 4 adalimumab levels [25%: 0-2.5 mg/mL; 50%: 2.6-5.4 mg/mL; 75%: 5.5-8.8mg/mL; 100%: 8.9-33.2mg/mL].Adalimumab levels were measured using an enzyme-linked immunosorbent assay in SERENE-UC 14 and a validated ligand-binding assay in SERENE-CD. 15The drug level stratification analyses included patients who received the higher dose of adalimumab, as well as those treated with the standard dose in SERENE-UC and SERENE-CD, to capture and assess data from patients with the widest possible range of drug exposure levels.Data for all these analyses are presented as box plots of the median [interquartile range], while the p-values were calculated using the baseline mean [standard deviation].All p-values were adjusted for multiple testing correction using the Benjamini-Hochberg method. 16To evaluate whether the results of this analysis were impacted by non-responders with inadequate drug levels, this analysis was repeated in only patients receiving the standard or high dose of adalimumab with adequate drug levels [>5 µg/mL] at Week 8 [SERENE-UC] and Week 12 [SERENE-CD].

SERENE-UC
The relationships between baseline TREM-1 gene expression and change from baseline levels, and neutrophil percentage as a proportion of the total leucocyte population, neutrophil absolute counts, and faecal calprotectin were assessed, grouped by endoscopic outcomes, and statistically tested using Pearson's correlation.Neutrophil percentage analyses were also performed stratified by prior CS use at baseline, given the possible impact of CS use on TREM-1 expression. 13The neutrophil analyses were performed using complete blood count levels, with neutrophils chosen because of their relatively high expression of TREM-1 compared with other leucocytes.All p-values reported throughout are nominal.Patients or the public were not involved in the design, conduct, reporting, or dissemination plans of our research.

Results
In SERENE-UC, whole blood TREM-  1 and 2] and SERENE-CD [Weeks 12 and 56; Supplementary Tables 3 and 4].The baseline demographics and clinical characteristics of both the UC and CD patient subsets analysed by endoscopic and clinical outcomes were generally representative of the overall study cohorts that have been reported previously. 14,15An exception to this was noted in SERENE-UC, where across all endpoints analysed, the proportion of male patients was higher in the overall population than in the RNA-seq subgroup [Supplementary Tables 1 and 2].Additionally, in SERENE-CD, a higher proportion of RNA-seq patients reported aminosalicylate use vs the overall population, and Clinical Disease Activity Index scores were slightly lower in the RNA-seq subgroup vs the overall population across all endpoints analysed [Supplementary Tables 3 and 4  91.8% of the 106 patients were White, 6.4% were Black or African American, and 1.8% were Asian.Finally, baseline CS use was associated with significantly lower baseline TREM-1 gene expression in SERENE-UC [p = 0.008], but not in SERENE-CD [p > 0.05].

Baseline TREM-1 expression levels by endoscopic and clinical outcomes
Among patients with UC who received the standard adalimumab dose, baseline TREM-1 gene expression did not differ in those who achieved endoscopic response at Week 8 vs non-responders at this time point; the same was true of endoscopic remitters vs non-remitters at Week 8 [Figure 1A].Significant but minor differences were observed based on endoscopic response [p = 0.0116] and endoscopic remission [p = 0.0484] at Week 52 [Figure 1A].Regarding clinical outcomes, baseline TREM-1 gene expression did not differ between patients with UC who achieved clinical response vs non-response at Week 8 or 52, or between clinical remitters vs non-remitters at Week 8; however, there was a significant but minor difference for clinical remission at Week 52 [p = 0.05; Figure 1B].The AUC values in patients with UC were low and ranged from 0.48 to 0.69.
In patients with UC, baseline soluble TREM-1 protein levels did not differ in endoscopic or clinical remitters or responders at Week 8 or 52 vs non-remitters and non-responders, respectively [p > 0.05].
In patients with CD, baseline TREM-1 gene expression did not differ in endoscopic remitters or responders at Week 12 or 56 vs non-remitters and non-responders, respectively, at these time points [Figure 2A].As with the UC population, there was also no significant difference in baseline TREM-1 gene expression in patients with CD who were clinical remitters vs non-remitters, or clinical responders vs non-responders, at either evaluation time point following standard adalimumab dosing [Figure 2B].The AUC values in patients with CD were low and ranged from 0.48 to 0.63.
In patients with CD, baseline soluble TREM-1 protein levels did not differ in endoscopic remitters or responders at Week 12 or 56 vs non-remitters and non-responders, or clinical remitters at Week 56 vs non-remitters [all p > 0.05]; however, there was a nominal difference between clinical responders and non-responders at Week 56 [p = 0.04].
Regardless of endoscopic or clinical outcome at any time point, TREM-1 gene expression generally decreased from baseline over time in both studies [Figures 3 and 4

Association of baseline TREM-1 gene expression with baseline neutrophil percentage and faecal calprotectin grouped by endoscopic outcomes
In patients treated with standard-dose adalimumab, baseline neutrophil percentage was highly positively correlated with baseline TREM-1 gene expression in responders and nonresponders, and remitters and non-remitters across endoscopic outcomes in patients with UC [Week 8: r = 0.68 for response vs r = 0.75 for non-response, and r = 0.74 for remitters vs r = 0.71 for non-remitters; Week 52: r = 0.66 for response vs r = 0.70 for non-response, and r = 0.67 for remitters vs r = 0.68 for non-remitters; Figure 5A] and CD [Week 12: r = 0.65 for response vs r = 0.48 for non-response, and r = 0.70 for remitters vs r = 0.48 for non-remitters; Week 56: r = 0.47 for response vs r = 0.58 for non-response, and r = 0.36 for remitters vs r = 0.58 for non-remitters; Figure 5B].All correlations were significant [p < 0.001] except for remitters at Week 56 in SERENE-CD [p = 0.06].No notable differences were observed between remitters vs non-remitters or responders vs non-responders.Similar results were seen when stratified further by patients with and without steroid exposure history, with no notable differences between groups.In addition, similar patterns were seen when analysing the relationships between absolute neutrophil counts instead of percentages and baseline TREM-1 expression, as well as change from baseline TREM-1 expression and neutrophil percentage [data not shown].
In contrast to the association seen with neutrophils, no strong association was observed between faecal calprotectin levels and baseline TREM-1 gene expression for any endoscopic outcome in patients with either UC [Week 8: r = −0.04

Discussion
The findings of this study demonstrate that baseline TREM-1 whole blood gene expression was not a predictor of clinical or endoscopic outcomes following adalimumab treatment in the Phase 3 SERENE studies in patients with moderately to severely active UC or CD.In SERENE-UC, baseline TREM-1 gene expression was weakly associated with endoscopic response and remission after 52 weeks of adalimumab treatment, but this was not sufficiently robust for TREM-1 to be considered a predictive biomarker of response to adalimumab.Similarly, baseline TREM-1 gene expression did not predict clinical outcomes in SERENE-UC, or clinical or endoscopic outcomes in SERENE-CD.Largely similar results were observed when analysed using TREM-1 soluble protein levels.
Stratification by per-quartile drug levels did not increase the ability of baseline TREM-1 gene expression to predict outcomes [including when only patients with adalimumab levels >5 µg/mL were analysed], suggesting that these results are not influenced by adalimumab dose.
The results of the current study are consistent with those of both the Personalising Anti-TNF Therapy in Crohn's disease study of patients with CD [NCT03088449] 17 and a small study in 33 children with IBD, in which baseline TREM-1 whole blood gene expression did not predict response to either adalimumab or infliximab. 11However, they contrast with the findings of other studies where TREM-1 appeared to be predictive of clinical and/or endoscopic outcomes in patients with IBD, albeit with opposing signals for low vs high baseline TREM-1 expression levels linked to response. 7,8,10he discrepancy between the results of our study and those of previous studies may be attributable to variations  7,11 Finally, the definitions of clinical and endoscopic outcomes and the timing of assessments differed between studies [Table 2].Regardless of clinical or endoscopic remission/nonremission and response/non-response, TREM-1 gene expression generally decreased over time following adalimumab treatment in both SERENE-UC and SERENE-CD.This suggests that TREM-1 could be a marker of anti-TNF treatment.Indeed, the link between downregulation of TREM-1 and response was specific to patients treated with anti-TNF agents in one of the studies by Verstockt et al. 7 It is possible that treatment of IBD with anti-TNF agents could reduce signalling through receptors such as TREM-1 and lead to general downregulation, even if this does not result in clinical or endoscopic response or remission.
Of note, a highly positive correlation was observed between baseline TREM-1 gene expression and the baseline percentage of neutrophils in both endoscopic remitters/non-remitters and responders/non-responders in both SERENE-UC [Weeks 8 and 52] and SERENE-CD [Weeks 12 and 56], except for remitters at Week 56 in SERENE-CD.This correlation was unaffected by steroid treatment history, and supports previous reports of TREM-1 activation modulating the function of neutrophils in models of IBD. 18While faecal calprotectin has been related to anti-TNF response and is usually monitored in clinical practice, 19,20 we did not find a strong correlation between TREM-1 expression and absolute baseline or change from baseline in faecal calprotectin levels.This is consistent with the findings of Verstockt et al. 7 Our analysis was limited in that it focused solely on TREM-1 expression, which has been suggested previously as a potential blood-based biomarker for anti-TNF response.Further analyses, using an unbiased, genome-wide approach as well as genetic studies at the individual level, may allow for identification of novel predictive biomarkers in whole blood and at the cellular level.However, findings generated from such a genome-wide approach would need confirmation in independent cohorts.Another limitation is that all patients in the subset of patients analysed from SERENE-UC and 92% of patients from SERENE-CD were White; this limited diversity means that the analysis may not be generalizable to other racial groups.Finally, dichotomous [response vs non-response] rather than continuous outcomes were analysed, which can lead to a loss of power for detecting a difference between groups; however, these outcomes were based on protocoldefined primary and secondary endpoints used for the primary analysis of each trial.
In summary, the results of this robust analysis in patients with moderately to severely active UC or CD from the large, randomized, Phase 3 SERENE studies demonstrate that baseline TREM-1 whole blood gene expression does not predict endoscopic or clinical outcomes following adalimumab treatment.Consequently, further exploration of potential predictors of response to anti-TNF therapy is needed to identify non-invasive biomarkers that could be used in clinical practice to allow personalized treatment and optimize treat-totarget management of IBD.

Figure 2 .
Figure 2. Baseline TREM-1 whole blood expression in patients with CD treated with the standard dose of adalimumab in SERENE-CD who had [A] endoscopic response/non-response and remission/non-remission at Weeks 12 and 56 and [B] clinical response/non-response and remission/nonremission at Weeks 12 and 56.a,b Data are median [interquartile range], with the p-values calculated using the baseline mean [standard deviation].a Definitions of endoscopic and clinical outcomes are listed in Table 2. b n = 106 at Week 12; n = 50 and n = 48 at Week 56 for endoscopic and clinical outcomes, respectively.CD, Crohn's disease; TREM-1, triggering receptor expressed on myeloid cells.

p
].In SERENE-UC, the decrease was particularly marked in endoscopic and clinical non-remitters [p < 0.001 at Week 8 among patients who were endoscopic or clinical non-remitters at Week 8 or endoscopic non-remitters at Week 52, and p < 0.05 at Week 8 among clinical non-remitters at Week 52; Figure3].By contrast, in SERENE-CD, the greatest decreases from baseline in TREM-1 gene expression were seen in endoscopic and clinical remitters and responders [p < 0.001 atWeek 12

Figure 3 .
Figure 3. TREM-1 whole blood expression over time in patients with UC treated with the standard dose of adalimumab in SERENE-UC who had [A] endoscopic response/non-response and remission/non-remission at Weeks 8 and 52 and [B] clinical response/non-response and remission/ non-remission at Weeks 8 and 52.a,b Data are median [interquartile range], with the p-values calculated using the baseline mean [standard deviation].a Definitions of endoscopic and clinical outcomes are listed in Table 2. b n = 95 at Week 8; n = 70 at Week 52.*p < 0.05.TREM-1, triggering receptor expressed on myeloid cells; UC, ulcerative colitis.

Figure 4 .
Figure 4. TREM-1 whole blood expression over time in patients with CD treated with the standard dose of adalimumab in SERENE-CD who had [A] endoscopic response/non-response and remission/non-remission at Weeks 12 and 56 and [B] clinical response/non-response and remission/nonremission at Weeks 12 and 56.a,b Data are median [interquartile range], with the p-values calculated using the baseline mean [standard deviation].a Definitions of endoscopic and clinical outcomes are listed in Table 2. b n = 106 at Week 12; n = 50 and n = 48 at Week 56 for endoscopic and clinical outcomes, respectively.*p < 0.05.CD, Crohn's disease; TREM-1, triggering receptor expressed on myeloid cells.

14 AFigure 5 .
Figure 5. Association of baseline TREM-1 whole blood expression with baseline neutrophil percentage by endoscopic response/non-response and remission/non-remission a in patients with UC and CD treated with the standard dose of adalimumab in [A] SERENE-UC and [B] SERENE-CD.All correlations were statistically significant [p < 0.05].a Definitions of endoscopic outcomes are listed in Table2.CD, Crohn's disease; CI, confidence interval; TREM-1, triggering receptor expressed on myeloid cells; UC, ulcerative colitis.

AFigure 6 .
Figure 6.Association of baseline TREM-1 whole blood expression with faecal calprotectin by endoscopic response/non-response and remission/non-remission a in patients with UC and CD treated with the standard dose of adalimumab in [A] SERENE-UC and [B] SERENE-CD.a Definitions of endoscopic outcomes are listed in Table2.*p < 0.05.CD, Crohn's disease; CI, confidence interval; TREM-1, triggering receptor expressed on myeloid cells; UC, ulcerative colitis.

Table 1 .
Adalimumab induction dosing regimens in SERENE-UC and SERENE-CD.

Table 2 .
Definitions of clinical and endoscopic response and remission used in the analysis of data from SERENE-UC and SERENE-CD, and in previously published studies of the ability of baseline whole blood TREM-1 expression to predict response to anti-TNF treatment in patients with IBD.

Endpoint SERENE-UC SERENE-CD Verstockt et al. 7 Prins et al. 9 Gaujoux et al. 10
11ntinuedin study designs and patient populations.For example, in the current study, samples were analysed from a relatively large number of patients from the SERENE randomized trials [95 and 106 patients with UC and CD, respectively].In contrast, previous studies analysed smaller samples of patients with IBD [54,722, 10 and 3311patients], generally treated in real-world clinical practice settings.Additionally, the methods used to measure whole blood TREM-1 gene expression differed between studies: we used RNA-seq to measure the full transcript, whereas real-time reversetranscription polymerase chain reaction was used in the studies byVerstocktand Salvador-Martin, with the full transcript and three TREM-1 isoforms being studied by Verstockt et al. [TREM-1-mb, TREM-1-sv, and TREM-1-x2].