Evaluation of Systemic Treatments of Small Intestinal Adenocarcinomas

Key Points Question Which treatment strategies for small intestinal adenocarcinomas (SIAs) are most beneficial for patient survival? Findings In this systematic review and meta-analysis of 57 studies, including 35 176 patients, adjuvant and palliative chemotherapy were associated with improved overall survival irrespective of tumor localization. Palliative fluoropyrimidine-oxaliplatin combinations were superior to other regimens for overall survival and progression-free survival, addition of bevacizumab to chemotherapy was associated with prolonged progression-free survival, and immunotherapy was associated with increased survival for patients with defective mismatch repair tumors. Meaning Current evidence supports the clinical use of fluoropyrimidine-oxaliplatin combinations as first-choice chemotherapy for SIAs, addition of bevacizumab to chemotherapy, and immunotherapy for patients with defective mismatch repair SIAs.


Introduction
Small intestinal adenocarcinomas (SIAs) are rare and poorly characterized.Their incidence varies from 1 to 3.5 per 100 000 inhabitants, despite the considerable size and high epithelial turnover rate of the small intestine. 1,24][5][6][7] This poor prognosis is partly due to diagnostic delay related to vague symptoms.
9][20][21] So far, these regimens have not been evaluated in dedicated randomized clinical trials of patients with SIAs.
Overall, the outcome of patients with SIAs is poor, molecular characteristics are unique, and the most effective systemic therapeutic regimens are largely unknown.We aimed to comprehensively summarize the available literature on survival benefit of systemic treatment in patients with SIAs.This effort indicates most beneficial treatment options and could inform future clinical trials to improve outcomes for patients with SIAs.

Study Design
This study was reported according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guidelines. 22Detailed descriptions of the methods are available in the eMethods in Supplement 1.
The primary objective was to assess the survival benefit of systemic therapies for locoregional and metastatic SIAs, 23 measured by OS.Secondary outcomes entailed progression-free survival (PFS), relapse-free survival (RFS), and cancer-specific survival (CSS) after systemic therapies for SIA, reported as hazard ratios (HRs) or median survival times in months (median OS [mOS] and median PFS [mPFS]).

Quality Assessment and Data Extraction
Consensus-based assessment of risk of bias, using the Cochrane Risk of Bias in Non-randomized Studies of Interventions (ROBINS-I) tool, 24 and data extraction were performed by 3 independent reviewers (T.d.B, I.N., and P.S.).Studies with high risk of bias were not excluded.Missing data were requested from the authors of the original studies.

Statistical Analysis
Meta-analyses were performed separately for the adjuvant and palliative setting with HRs for OS, RFS, PFS, and CSS and with mOS and mPFS data.For each meta-analysis, subsets of the included studies were selected depending on reported data and line of therapy.We performed randomeffects, inverse variance, pairwise meta-analyses.To compare 4 first-line regimens, network metaanalyses (NMAs) were performed using the GeMTC platform. 25terogeneity was assessed with the Higgins I 2 index and the Cochran Q test; heterogeneity was considered present if an I 2 was 50.0% or greater or a Cochrane Q test P Յ .05.Studies that contributed to heterogeneity in a meta-analysis were excluded in sensitivity analyses to assess their effect on the estimated outcome.In addition, subgroup analyses and metaregression for geography, risk of bias, publication period, adjustment of HRs, and stage and line of therapy were run to identify potential causes for heterogeneity.Nonheterogeneous effect estimates were reported.Publication bias was assessed with funnel plots and the Egger test. 26l analyses were performed with R software, version 4.0.5 (R Foundation for Statistical Computing) and the GeMTC platform, version 1.0-1. 25Hypothesis tests were 2-sided, with a significance level of P < .05.

Selection of Eligible Studies
The search yielded 4353 studies, of which 116 were eligible for further review (eFigure 1 in Supplement 1).After full-text revision, 57 studies were included, with a total of 35 176 patients.
None of these variables were significantly associated with the estimated benefit from palliative chemotherapy.
b Combinations with oxaliplatin, carboplatin, and cisplatin.
d Predominantly platinum combination and fluoropyrimidine monotherapy.
f Fluoropyrimidine monotherapy and platinum combination.Response rates to palliative chemotherapy were consistent with our previous analyses (eTable 2 in Supplement 1).Overall response rates for fluoropyrimidine-oxaliplatin varied from 34.0% to 56.0%. 28,32,59,72,75For fluoropyrimidine-irinotecan, the overall response rates ranged from 9.0% to 55.0%, 48,59,61,72,75 and for fluoropyrimidine-cisplatin, the overall response rates ranged from 31.0% to 38.0%. 59,72,75For triplet regimens, overall response rates of 83.0% (folinic acid, fluorouracil, and irinotecan-oxaliplatin) and 38.0% (capecitabine, irinotecan, and oxaliplatin) were reported. 29,61terogeneity, Publication Bias, and Certainty of Evidence Eleven of our 31 pairwise meta-analyses pooling HRs and 7 of our 15 meta-analyses with medians were heterogeneous.eTables 8 and 9 in Supplement 1 demonstrate correction of heterogeneity in sensitivity analyses.Heterogeneity was partially explained by publication date and adjustment of HRs in the palliative setting (eTable 7 in Supplement 1), differences in cohort sizes, and treatment regimens.Specifically, inclusion of cisplatin doublets caused heterogeneity in the meta-analyses of platinum combinations vs other regimens.After exclusion of outlier studies, heterogeneity was resolved.
In our NMAs, heterogeneity was limited, as observed with the random-effects SDs (OS: SD, 0.30; range, 0.02-0.76;PFS: SD, 0.36; range, 0.02-0.76),which were both smaller than the effect sizes in the models.No outliers or influential studies were detected when assessing residual deviance and leverage statistics.
Of our 46 meta-analyses, 2 were found to have publication bias (eFigures 15-20 in Supplement 1).These were the meta-analyses on the OS benefit of second-line chemotherapy compared with no treatment and on mPFS after second-and third-line palliative chemotherapy.The results of these analyses should be carefully interpreted.
Certainty of evidence of all meta-analyses was assessed using the Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) tool (eTables 10 and 11 in Supplement 1). 80Eleven of our meta-analyses scored very low, 12 scored low, 21 scored moderate, and 2 scored high.

Prospective Evaluation of Targeted Agents and Immunotherapy
One study 20 prospectively assessed CAPOX with bevacizumab in 23 patients with SIA and 7 patients with AAC and found an mOS of 12.9 months, an mPFS of 8.7 months, and an overall response rate of 48.3%, suggesting that CAPOX with bevacizumab is an active regimen for advanced SIA.On the contrary, panitumumab did not improve survival in 9 patients with RAS (OMIM 606614) wild-type tumors (including 1 patient with AAC), with an mOS of 5.7 months and an mPFS of 2.4 months. 81garding immunotherapy, the anti-programmed cell death ligand 1 antibody avelumab was tested in 8 patients (including 3 patients with AAC), without selection for dMMR status, and revealed an mPFS of 8.0 months. 82Pembrolizumab, targeting programmed cell death 1, was administered to 40 patients with SIAs and found an overall response rate of 50% in the subgroup of 4 patients with dMMR tumors. 79

Discussion
This meta-analysis aimed to summarize prospective and retrospective data on the survival benefit of systemic therapies for patients with SIAs.Our meta-analyses show benefit from adjuvant chemotherapy regardless of localization and increasing by incremental disease stage.Furthermore, chemotherapy extends OS and PFS in the palliative setting.Consistently, first-line fluoropyrimidineoxaliplatin was most favorable for survival in both pairwise meta-analyses and NMAs.Addition of bevacizumab to first-line chemotherapy significantly improved PFS compared with chemotherapy alone.Finally, immunotherapy seemed to prolong survival for patients with dMMR tumors.
Studies to refine clinical applicability of adjuvant chemotherapy are needed for several reasons.
First, selection of patients with SIAs for adjuvant treatment remains unclear.For CRC, high-risk patients with stage II or stage III disease are eligible, but it is unknown whether these selection criteria also apply to patients with SIAs.In addition, patients with duodenal adenocarcinomas are at higher risk of being unfit for adjuvant chemotherapy after Whipple surgery compared with patients with distal adenocarcinomas.Second, specific regimens need to be compared in randomized clinical trials to determine most effective therapies.For this purpose, a first international phase 3 trial was set up (Global BALLAD [A Trial to Evaluate the Potential Benefit of Adjuvant Chemotherapy for Small Bowel Adenocarcinoma]) to compare observation alone, monotherapy with fluoropyrimidines, and FOLFOX for patients with stage I to III SIAs. 84Simultaneously, a phase 3 trial has been initiated in Japan (J-BALLAD) in which observation alone will be compared with CAPOX. 85Third, tailoring therapy to molecular tumor traits requires further attention, for example, (neo-)adjuvant immunotherapy for patients with localized dMMR SIAs.
Only a few studies in patients with metastatic SIAs assessed survival benefit of palliative chemotherapy.In most studies, 7,28,32,59,72,75,86 fluoropyrimidine-oxaliplatin was superior to other regimens in terms of survival and response rates.This finding is perhaps not unexpected because this regimen is also highly effective in other gastrointestinal cancers. 87,88For cisplatin combinations, our study suggested a smaller survival benefit compared with other regimens.The 2 studies 59,74 that demonstrated a higher mOS and mPFS for cisplatin compared with oxaliplatin combinations included mainly duodenal adenocarcinomas (80%), which might be sensitive to cisplatin-based treatment due to their anatomical proximity to the pancreas and stomach. 89,90rgeted agents are rarely used in patients with SIAs because of unclear efficacy.Our results do not allow for any definitive conclusions on this matter.However, addition of bevacizumab to chemotherapy appeared to be more beneficial for survival than chemotherapy alone.This finding is in line with trials performed in metastatic CRC. 91For anti-epidermal growth factor receptor therapy, hardly any data are available.The available data do not support clinical activity, even in patients with RAS wild-type tumors. 63,81This finding might be related to the reduced activity seen in the right side of the colon, which develops from the same embryonic structure as the small intestine (midgut).
Regarding immunotherapy, evidence shows its activity in varying solid dMMR cancers, with response rates ranging from 38.0% to 50.0% and even long-term survival. 21,79,92,93These data indicate that immune checkpoint inhibition should be the preferred treatment choice in dMMR metastatic SIAs over bevacizumab.

Limitations
Our meta-analysis is affected by several limitations.Unfortunately, most studies 1,3,4,17,19, were nonrandomized cohort studies. Hence the outcomes of some of our meta-analyses are subject to selection bias and confounding by prognostic factors.The calculated survival benefit of chemotherapy administration, compared with no chemotherapy, is likely overestimated because of routine selection of fitter patients for systemic therapy.In our view, this limitation plays a lesser role in the meta-analyses that compare different treatment regimens because selection bias stemming from patient performance or disease burden is likely more limited in this case.Furthermore, with subgroup and metaregression analyses, we confirmed that risk of bias did not significantly impact the results of our meta-analyses.In addition, the use of both adjusted and unadjusted HRs did not significantly impact outcomes, as shown in our metaregression analyses.However, differences in adjustment variables could introduce bias, although the variables adjusted for were relatively consistent across studies.Furthermore, the scarcity of data precluded most pairwise regimen comparisons.Similarly, the numbers of eligible studies for the NMAs were limited.Therefore, our MEDLINE and Embase were searched for studies published between January 1, 2005, and June 1, 2022, on the survival benefit of systemic therapies for patients with SIAs.Eligible studies contained HRs, survival times, or response rates after systemic therapies.Studies on ampullary adenocarcinomas (AACs), surgery, or radiotherapy were excluded.Study selection was performed by 3 reviewers independently (T.d.B., I.N., and P.S.).

Figure
Figure 1.Overall Survival and Relapse-Free Survival Benefit of Adjuvant Chemotherapy per Stage Group and per Localization

Figure
Figure 2. Association of Palliative Chemotherapy With Overall Survival Benefit for the Total Subset, in the First-and Second-line Settings, and for Duodenal Adenocarcinomas Alone and With Overall Survival and Progression-Free Survival After Palliative Chemotherapy for Duodenal Adenocarcinomas Compared With Adenocarcinomas From Jejunum and Ileum

Figure 4 .
Figure 4. Median Survival in Patients With Metastatic Small Intestinal Adenocarcinomas Doublets seemed to be associated with improved PFS compared with singlets but not with prolonged OS.Triplets were not associated with improved PFS compared with doublets.
2. Association of Palliative Chemotherapy With Overall Survival Benefit for the Total Subset, in the First-and Second-line Settings, and for Duodenal Adenocarcinomas Alone and With Overall Survival and Progression-Free Survival After Palliative Chemotherapy for Duodenal Adenocarcinomas Compared With Adenocarcinomas From Jejunum and Ileum Figure 3.Comparison of Associations of First-line Palliative Chemotherapy Regimens With Overall Survival and Progression-Free Survival JAMA Network Open.2023;6(2):e230631.doi:10.1001/jamanetworkopen.2023.0631(Reprinted) February 24, 2023 6/16 Downloaded From: https://jamanetwork.com/ on 09/28/2023 have been inconclusive with regard to benefit from adjuvant chemotherapy.Our meta-analyses are in line with large retrospective cohort Downloaded From: https://jamanetwork.com/ on 09/28/2023 studies Overview of studies describing adjuvant chemotherapy eTable 2. Overview of studies describing palliative chemotherapy eTable 3. Risk of bias assessment using the Cochrane ROBINS-I tool eTable 4. Subgroup analyses for adjuvant chemotherapy versus no chemotherapy eTable 5. Metaregression for adjuvant chemotherapy versus no chemotherapy eTable 6. Subgroup analyses for palliative chemotherapy versus no chemotherapy eTable 7. Metaregression for palliative chemotherapy versus no chemotherapy eTable 8. Sensitivity analyses for meta-analyses with hazard ratios eTable 9. Sensitivity analyses for meta-analyses with median survival data eTable 10.Certainty of evidence score for meta-analyses with hazard ratios eTable 11.Certainty of evidence score for meta-analyses with median survival data Forest plots eFigure 12. Forest plots eFigure 13.Network meta-analyses for first-line regimens eFigure 14.Comparison of pairwise and network meta-analyses eFigure 15.Evaluation of publication bias eFigure 16.Evaluation of publication bias eFigure 17.Evaluation of publication bias eFigure 18. Evaluation of publication bias eFigure 19.Evaluation of publication bias eFigure 20.Evaluation of publication bias eReferences JAMA Network Open.2023;6(2):e230631.doi:10.1001/jamanetworkopen.2023.0631(Reprinted) February 24, 2023 15/16 Downloaded From: https://jamanetwork.com/ on 09/28/2023 eFigure 11.