NCCTG N0543 (Alliance): A Phase II Trial of Pharmacogenetic-Based Dosing of Irinotecan, Oxaliplatin, and Capecitabine as First-Line Therapy for Advanced Small Bowel Adenocarcinoma

Abstract

Background

Oxaliplatin in combination with either 5-fluorouracil or capecitabine is commonly used as first line therapy for small bowel adenocarcinoma. The addition of irinotecan improves survival in other gastrointestinal tumors but at the cost of hematologic toxicity. We performed a phase II, cooperative group study (NCCTG N0543, Alliance) using genotype-dosed capecitabine, irinotecan, and oxaliplatin (gCAPIRINOX), with dosing assigned based on UGT1A1 genotype in order to 1) test whether the addition of irinotecan would improve outcomes and whether 2) UGT1A1 genotype-based dosing could optimize tolerability.

Methods

Previously untreated patients with advanced small bowel adenocarcinoma received irinotecan (day 1), oxaliplatin (day 1), and capecitabine (days 2–15) in a 21-day cycle and were dosed with gCAPIRINOX according to UGT1A1*28 genotypes (6/6, 6/7, and 7/7).

Results

33 pts (17-6/6, 10-6/7, 6-7/7) were enrolled [73% male, mean age 64 (range 41–77)] from October 2007 to November 2013. Location of primary tumor included: duodenum (58%), jejunum (30%), and ileum (9%). The regimen yielded a confirmed response rate of 37.5% (95% CI 21–56) with median progression-free survival of 8.9 months and median overall survival of 13.4 months. Neither hematologic toxicity (grade 3+ 52.9%, 30.0%, 33.3%, respectively) nor tumor response rate (41.2% 33%, 33%) differed significantly by UGT1A1 genotype (6/6, 6/7, and 7/7 groups).

Conclusions

UGT1A1 genotype-directed dosing (gCAPIRINOX) is feasible with favorable rates of hematologic toxicity compared with prior three drug studies in unselected patients. Larger studies would be needed to determine comparability to CapeOx alone or if response/toxicity differs among genotypes.

Introduction

Small bowel adenocarcinoma (SBA) is a rare malignancy that is often diagnosed in its late stages, with metastatic disease. Due to its rarity, it is historically understudied, and as such there is no clear, consistent standard of care. To date, small prospective clinical studies have evaluated chemotherapy for SBA, the first of which examined the use of mitomycin C, adriamycin, and 5-fluorouracil (FAM) with minimal therapeutic effect, which likely explains why the study was performed in the early 1980s but not reported until 2005.¹ More recently, oxaliplatin based therapies have been demonstrated to have activity and have emerged as a potential standard of care in the front line setting. Overman et al. demonstrated a high response rate (50%) in 30 patients with combination therapy with oxaliplatin and capecitabine (CapeOx), although the trial also included 12 patients (40%) with ampullary carcinoma, which may not necessarily behave in a similar clinical fashion.² Median progression-free survival (PFS) and overall survival (OS) were 9.4 months and 15.5 months, respectively. This study showed that small bowel carcinoma indeed is a chemosensitive disease with response rates and activity of chemotherapy, namely 5-FU and oxaliplatin-based therapy, similar to other gastrointestinal malignancies such as gastric³ or colorectal cancer⁴. Xiang et al performed a study using FOLFOX in SBA showed similar response rates (48%), median progression-free survival (7.8 months), and median overall survival (15.2 months) to the CapeOx study⁵, confirming the activity of an oxaliplatin-based doublet in SBA.

In a phase I study of irinotecan, oxaliplatin, 5-fluorouracil, and leucovorin, 2 of 5 patients with small bowel adenocarcinoma showed partial responses to the regimen.⁶ However, the recommended phase II schedule proved too toxic in a cooperative group colorectal cancer study.⁷ A triplet combination of oxaliplatin, irinotecan, and 5-fluorouracil has shown survival benefit in pancreatic and colorectal cancers.^8–11 UGT1A1, the metabolic enzyme involved in the clearance of SN-38, the active metabolite of irinotecan, has common polymorphisms in the number of TA repeats (6 vs 7), and affects the toxicity profile of irinotecan, including risk for neutropenia, diarrhea, and efficacy.^12–17 A 7/7 genotype is perhaps best known as the most common cause of Gilbert’s syndrome, a generally asymptomatic indirect hyperbilirubinemia.¹⁸ Genotype-driven dosing for irinotecan has been successfully accomplished in early phase trials.^{17, 19} A phase I trial with oxaliplatin, irinotecan, and capecitabine (gCAPIRINOX) was performed at Mayo Clinic using UGT1A1 genotype performed prospectively in parallel cohorts to determine the appropriate dose for the three common genotypes, with the identification of higher tolerable doses in the 6/6 compared to the 6/7 and 7/7 with comparable total SN38 exposure.²⁰ Using the doses identified in that study for each genotype (6/6, 6/7, 7/7), the North Central Cancer Treatment Group (NCCTG) performed a pharmacogenetic-based phase II study (N0543) in advanced small bowel adenocarcinoma patients. The NCCTG is now a part of the Alliance for Clinical Trials in Oncology.

Patients and Methods

Inclusion criteria were unresectable locally advanced or metastatic small bowel adenocarcinoma, measurable disease, life expectancy ≥ 12 weeks, ECOG performance score of 0, 1, or 2, age ≥ 18 years, at least 4 weeks since major surgery and at least 2 weeks since radiation treatment, adequate blood counts and organ function. Key exclusion criteria included prior therapy for advanced small bowel cancer (although adjuvant 5-FU and leucovorin were allowed if it had ended 3 more or months prior to registration), serious intercurrent illness, or current other malignancy.

Prior to registration, each participant signed an IRB-approved, protocol-specific informed consent in accordance with federal and institutional guidelines. After registration, patients had germline blood DNA testing for UGT1A1 genotype by polymerase chain reaction (PCR) with fragment analysis by capillary gel electrophoresis in a CLIA-approved laboratory (Mayo Medical Labs). Patients were assigned to treatment cohorts based on the number of TA repeats at the *28 locus, at doses based on the phase I study as follows: 6/6 genotype (Arm A)– irinotecan 150 mg/m2 (d1), oxaliplatin 100 mg/m2 (d1), and capecitabine 1600 mg/m2 divided twice daily (d2-15) in a 21 day cycle; 6/7 genotype (Arm B) – irinotecan 150 mg/m2 (d1), oxaliplatin 85 mg/m2 (d1), and capecitabine 400 mg/m2 divided twice daily (d2-15); and 7/7 patients (Arm C) – irinotecan 75 mg/m2 (d1), oxaliplatin 85 mg/m2 (d1), and capecitabine 400 mg/m2 divided twice daily (d2-15). Patients were then treated until progression or intolerance, and reimaged at 8-week intervals or as clinically indicated.

For celiac disease screening, Tissue Transglutamase IgA were assessed by enzyme-linked immunosorbent assay (ELISA) on baseline serum levels in a CLIA laboratory (Mayo Medical Labs). Results were reported to the principal investigator (RRM), and positive (>10 U/mL) results were reported to the treating physician. Mismatch repair status was determined by immunohistochemistry for detected absence of staining for any of MLH1, MSH2, MSH6, and PMS2 on tumor tissue from prior surgery or biopsies, and interpreted by an expert pathologist (TCS).

Statistical Methods

The primary objective of this phase II study was to report the proportion of patients with a confirmed tumor response across all treatment arms combined. The genotype-based arms were pooled for toxicity and efficacy analyses, although secondary analyses explored potential differences among arms. All patients who met the eligibility criteria, signed a consent form, and began treatment were included in the primary analysis.

A two-stage phase II 3-outcome design²¹ was used to test whether there was sufficient evidence to determine if the confirmed response rate was at least 30%, a rate the study team viewed as clinically promising compared to the FAM data, the only available benchmark at the time the study was initiated, versus at most 10%, a rate the study team viewed as not clinically promising. When the FOLFOX data of Overman et al² became available, a decision was made to complete accrual rather than redesign the study on new endpoint criteria, given the rarity of the tumor type and paucity of data in a cooperative group setting. With 33 patients, this trial carried 90% power to detect a tumor response rate of 30% with a 0.04 level of significance. A stage 1 analysis was to be performed after the first 16 evaluable patients were enrolled. If 1 or fewer responses were observed in this initial cohort, further accrual to the trial was to have been abandoned. If 2 or more responses were observed, the study would proceed to full accrual, where 7 or more confirmed responses in the first 33 eligible patients would be considered promising and worthy of further study. A confidence interval for the percentage of patients with a tumor response was calculated using the exact binomial method.

Secondary endpoints include descriptive summaries of adverse events, progression-free survival, and overall survival. In addition, exploratory secondary analyses estimated the efficacy (response and survival) separately by genotype (arm), mismatch repair (MMR) status, and tumor location. Adverse events (AEs) are summarized in a tabular manner as the maximum grade for each AE and patient, where those AEs at least possibly related to treatment are reported. Progression-free survival (PFS) was defined as the time from randomization to the first of either disease progression or death from any cause. Overall survival was defined as the time from study registration to death from any cause. Time-to-event distributions were estimated with the Kaplan-Meier method²², where the log-rank test was used to compare these Kaplan-Meier estimates for different subgroups of interest (i.e., arm, MMR status). The Fisher’s exact test was used for the associations of categorical variables and the Kruskal-Wallis test was used to assess the associations between continuous variables and categorical variables with 3 or more categories. All statistical tests for secondary endpoints were two-sided, where a p-value < 0.05 was considered statistically significant. Data collection was conducted by the Alliance Statistics and Data Center. All analyses were performed by Alliance statisticians using SAS version 9.4 software (SAS Institute, Cary, NC).

The principal investigator and the study statistician reviewed the study periodically (at least twice per year, per the Data and Safety Monitoring Board (DSMB) schedule) to identify accrual, toxicity, and any endpoint problems that might be developing.

Results

A total of 33 patients were enrolled on to this trial from October 2007 to November 2013 across 15 NCCTG/Alliance treating locations. One patient was later found to be a major treatment violation during cycle 1, since they received double the expected dose, so this patient was excluded from all efficacy analyses. All 33 patients were included in the demographics, treatment data, and adverse event data summaries.

Baseline patient characteristics are summarized in Table 1 overall and by genotype (arm). Median time from preregistration (including genotyping if not already completed) to first treatment was 10 days (range 0–52 days). No significant differences were found between the arms. The median age was 64 (range: 41–77), 24 (73%) were male, and 30 (91%) had an ECOG PS of 0 or 1. The most common tumor location was the duodenum [19 (58%)], followed by the jejunum [10 (30%)], and ileum [3 (9%)]. Other baseline characteristics are summarized in Table 1. Despite prior reports of celiac disease being present in up to 8% of patients with SBA²³, 0 of 33 patients tested had positive serology for tissue transglutamase IgA.

Table 1.

Patient Characteristics of Participants

	Arm A (6/6 Mutation, N=17)	Arm B (6/7 Mutation, N=10)	Arm C (7/7 Mutation, N=6)	Total (N=33)	P-value
Age					0.661
Median (Range)	64.0 (41.0–75.0)	66.5 (42.0–77.0)	62.5 (43.0–74.0)	64.0 (41.0–77.0)
Gender					0.232
Male	12 (70.6%)	9 (90.0%)	3 (50.0%)	24 (72.7%)
ECOG Performance Score					0.952
0	9 (52.9%)	5 (50.0%)	3 (50.0%)	17 (51.5%)
1	7 (41.2%)	4 (40.0%)	2 (33.3%)	13 (39.4%)
2	1 (5.9%)	1 (10.0%)	1 (16.7%)	3 (9.1%)
Race					0.422
White	16 (94.1%)	10 (100.0%)	5 (83.3%)	31 (93.9%)
Black or African American	1 (5.9%)	0 (0.0%)	1 (16.7%)	2 (6.1%)
Ethnicity					1.002
Not Hispanic or Latino	16 (94.1%)	10 (100.0%)	6 (100.0%)	32 (97.0%)
Unknown: Patient is unsure of their ethnicity	1 (5.9%)	0 (0.0%)	0 (0.0%)	1 (3.0%)
Primary Tumor Location					0.392
Duodenum	9 (52.9%)	7 (70.0%)	3 (50.0%)	19 (57.6%)
Jejunum	7 (41.2%)	2 (20.0%)	1 (16.7%)	10 (30.3%)
Ileum	1 (5.9%)	1 (10.0%)	1 (16.7%)	3 (9.1%)
Cannot Discern	0 (0.0%)	0 (0.0%)	1 (16.7%)	1 (3.0%)
Prior Radiotherapy					0.682
Yes	2 (11.8%)	0 (0.0%)	0 (0.0%)	2 (6.1%)
No	15 (88.2%)	10 (100.0%)	6 (100.0%)	31 (93.9%)
Prior Chemotherapy					0.422
Yes	1 (5.9%)	0 (0.0%)	1 (16.7%)	2 (6.1%)
No	16 (94.1%)	10 (100.0%)	5 (83.3%)	31 (93.9%)
MMR Status					1.002
dMMR	2 (25.0%)	0 (0.0%)	1 (33.3%)	3 (21.4%)
pMMR	6 (75.0%)	3 (100.0%)	2 (66.7%)	11 (78.6%)

¹Kruskal-Wallis p-value;

²Fisher’s exact p-value

Treatment Data

Across all patients, a median of 6 cycles of therapy was given (range: 1–32). All patients went off study treatment. Reasons for going off study treatment were similar between the 3 genotype groups (p= 0.10). Most patients ended treatment due to progressive disease (58%). Additional reasons for ending treatment consisted of adverse events (18%), patient refusal (6%), and other miscellaneous reasons.

Efficacy Data

Thirty-two patients were evaluable for efficacy measures. Of the first 16 eligible patients, 7 patients had a confirmed tumor response, which allowed continuation to full accrual. In all 32 evaluable patients, there were 12 confirmed tumor responses (10 partial responses, 2 complete responses) which met the protocol-defined criteria for success with an overall confirmed response rate of 38% (95% CI: 21–56%). Response rates were similar between arms and tumor locations (p=1.00; Table 2). Sufficient tissue for MMR analysis was available on 14 patients. We saw a higher observed response rate in patients with MMR-deficient tumors as compared to patients with MMR-proficient tumors (67% vs. 27%), but this failed to reach significance possibly due to small numbers (p=0.51). Across all patients, 14 had a best response of stable disease (44%). The overall clinical benefit rate (SD + Response) was 26/32 (81%). Waterfall plotting (Figure 1) shows the majority of patients had some shrinkage of tumor in each treatment arm, although not necessarily meeting the criteria for a confirmed partial or complete response, and activity was seen regardless of treatment arm, MMR status, or primary site (Supplemental figures s1–3)

Table 2.

Confirmed Responses Overall and By other Subgroups of Interest

	Success/Total (%)	95% Binomial Exact CI	Fisher’s Exact P-value
Confirmed Response
Pooled across all patients	12/32 (37.5)	21.1–56.3
Arm			1.00
A (6/6 genotype)	7/17 (41.2)	18.4–67.1
B (6/7 genotype)	3/9 (33.3)	7.5–70.1
C (7/7 genotype)	2/6 (33.3)	4.3–77.7
Tumor Location			1.00
Duodenum	7/18 (38.9)	17.3–64.3
Ileum	1/3 (33.3)	0.8–90.6
Jejunum	4/10 (40.0)	12.2–73.8
MMR Status			0.51
dMMR	2/3 (66.7)	9.4–99.2
pMMR	3/11 (27.3)	6.0–61.0

Figure 1.

Waterfall plot of maximal response for each small bowel adenocarcinoma patient treated with pharmacogenetic-based dosing of capecitabine, irinotecan, and oxalpliatin (gCAPIRINOX) by Treatment Arm. Patients with UGT1A1*28 6/6 genotype were assigned to irinotecan 150 mg/m2 (d1), oxaliplatin 100 mg/m2 (d1), and capecitabine 1600 mg/m2 divided twice daily (d2-15) in a 21 day cycle (Arm A). Patients with 6/7 genotype were treated with irinotecan 150 mg/m2 (d1), oxaliplatin 85 mg/m2 (d1), and capecitabine 400 mg/m2 divided twice daily (d2-15, Arm B). For 7/7 patients, irinotecan was dosed at 75 mg/m2 (d1), oxaliplatin 85 mg/m2 (d1), and capecitabine 400 mg/m2 divided twice daily (d2-15, Arm C).

Of the 32 patients evaluable for survival, 31 had passed away at the time of this report, with a follow-up of 59.4 months in the 1 patient still alive. The median overall survival across all patients was 13.4 months [95% confidence interval (CI) 10.5–18.1 months] which was similar across arms (range 12.1–15.7 months, p=0.52) and primary tumor sites (range 12.4–16.8 months, p=0.63) (Supplemental Figure s4). The median progression-free survival was 8.9 months (95% CI 4.7–10.8 months) and was also similar across arms (p=0.38) and tumor sites (p=0.59)) (Supplemental Figure s5). Finally, for those tumors with tissue available for determination of MMR status, dMMR (N=3) vs pMMR (N=11) was not found to be prognostic for OS (12.2 vs 10.0 months, p= 0.44) or PFS (10.6 vs 7.1 months, p=0.61, Supplemental figure s6), although the analyses were underpowered to detect a small or even moderate difference in outcome.

Adverse Events

All 33 patients were evaluable for adverse events (Table 3). Of these 33 patients, 26 (79%) had a grade 3 or worse AE, at least possibly related to treatment. Of all patients, 6 (18%) had a grade 4 AE and one patient had a grade 5 AE (suicide, unrelated to treatment). No significant differences were found for the overall AE rates between arms. Commonly occurring grade 3/4 adverse events (5 or more) consisted of neutropenia (9; 27%), leukopenia (6; 18%), dehydration (5; 15%), diarrhea (7; 21%), nausea (8; 24%), and vomiting (6; 18%). The numbers were too small for statistical comparisons between arms to be meaningful for the individual adverse events. Any grade diarrhea was reported in 88% of Arm A (6/6) patients (53% grade 1, 6% grade 2, 29% grade 3), 90% of Arm B (6/7) patients (50%, 20%, 20%, respectively), and 50% Arm C (7/7 patients) (17%, 33%, 0%). There were no reports of grade 4/5 diarrhea in any arm. Grade 3/4 neutropenia was noted in 29% of Arm A, 30% of Arm B, and 17% of Arm C patients, respectively. Grade 3 Febrile neutropenia was reported in one Arm A patient (6%) and one Arm B patient (10%) patient, respectively. Detailed adverse event data are noted in Supplemental Table s1.

Table 3.

Frequency of Adverse Events (AEs) by Treatment Arm (N = 33)

	Arm	N	%	P-value
Grade 3+ AE	All	26	78.8
	A	14	82.4	0.841
	B	7	70.0
	C	5	83.3
Grade 4+ AE	All	6	18.2
	A	4	23.5	0.831
	B	1	10.0
	C	1	16.7
Grade 3+ Hematologic AE	All	14	42.4
	A	9	52.9	0.571
	B	3	30.0
	C	2	33.3
Grade 4+ Hematologic AE	All	4	12.1
	A	2	11.8	1.001
	B	1	10.0
	C	1	16.7
Grade 3+ Non-Hematologic AE	All	22	66.7
	A	12	70.6	0.881
	B	6	60.0
	C	4	66.7
Grade 4+ Non-Hematologic AE	All	2	6.1
	A	2	11.8	0.661

All adverse events possibly, probably, or definitely related to treatment

¹Fisher’s exact test

^*Patients with UGT1A1*28 6/6 genotype were assigned to irinotecan 150 mg/m2 (d1), oxaliplatin 100 mg/m2 (d1), and capecitabine 1600 mg/m2 divided twice daily (d2-15) in a 21 day cycle (Arm A). Patients with 6/7 genotype were treated with irinotecan 150 mg/m2 (d1), oxaliplatin 85 mg/m2 (d1), and capecitabine 400 mg/m2 divided twice daily (d2-15, Arm B). For 7/7 patients, irinotecan was dosed at 75 mg/m2 (d1), oxaliplatin 85 mg/m2 (d1), and capecitabine 400 mg/m2 divided twice daily (d2-15, Arm C).

Discussion

While there are no randomized controlled clinical trials establishing a clear standard of care for SBA, several phase II trials have provided evidence of clinical activity of chemotherapeutic regimens for this disease. Given the similar progression-free and overall survival in our study compared to that of CapeOx² and FOLFOX⁵, it may be that the addition of irinotecan does not meaningfully add to the activity of CapeOx alone. However, such a view would not account for the differing clinical outcomes for single center trials versus a cooperative group trial, such as this one. One of the more salient aspects of our study is the reliance on genotyping to determining dosing, as this represents the first pharmacogenetically dosed clinical trial for small bowel adenocarcinoma. Importantly, we were successful in implementing genotyping within the context of a multi-institutional trial. Moreover, this approach appears to account for our relatively favorable adverse event profile, and the relative consistency of the outcomes across arms supports this approach.

In the study poluation, gCAPIRINOX was well tolerated overall, and it appears the triplet regimen dosed by UGT1A1 genotype may have an improved safety profile compared with what has been reported with FOLFIRINOX and FOLFOXIRI, with the appropriate caveats for comparison across studies.^{8, 10, 24} As potentially expected given the impact of UGT1A1 genotype primarily on neutropenia¹⁴, we identified a 27% rate of grade 3/4 neutropenia compared with 46%–59% for FOLFIRINOX/FOLFOXIRI (Supplemental Table s2) ; although diarrhea was comparable to slightly higher at 21% vs 13–20%%. There did not appear to be notable differences in toxicity rates among the 3 arms of our study (Table 3, supplementary Table 1), although the small numbers of patients preclude identification of moderate differences. For 7/7 genotype patients alone (Arm A), grade 3/4 neutropenia occurred in 29%, with 6% febrile neutropenia, while diarrhea was higher at 29%, perhaps the latter partially a result of the higher dosing of capecitabine. Combination of capecitabine, oxaliplatin, and irinotecan have not been published, aside from an unsuccessful phase I of CAPIRINOX in combination with radiation for rectal cancer.²⁵ A neoadjuvant pancreatic cancer study utilizing CAPOXIRI (NCT01760252) is ongoing. Ideally, in our study, the doses of irinotecan alone would have been assigned based on UGT1A1 genotype, but the prior phase I study determined the doses assigned.

There may exist a subset of patients with dMMR tumors who may be candidates for immune checkpoint inhibitor/anti-PD-1 therapy based on the recent report of pembrolizumab therapy showing responses in patients with dMMR tumors with non-colorectal diagnosis. Among the trials reported were 2 SBA patients, although their outcomes are not specifically reported.²⁶ In our study, 21% of patients with tissue available to be tested were found to be dMMR by immunohistochemistry, which appears to be a notably higher proportion than the 5% seen in metastatic colon cancer.^{27, 28} Thus, though SBA is a rare disease compared to colorectal cancer, there is likely notable opportunity for immune checkpoint blockade in SBA.

There may be biological differences in small bowel adenocarcinoma according to site of the primary tumor. Much like colorectal cancer^{29, 30}, there appears to be genetic heterogeneity among sites of small bowel adenocarcinoma primary tumor origination, with MSI-H tumors appearing more often in more proximal tumors and KRAS mutations are more common in duodenal cancers (57%) vs jejunum (29%) and ileum (14%)).³¹ Her-2 expression levels appear to be lowest in proximal tumors, and highest in ileal tumors. These findings appear to argue against the commonly held opinion that proximal tumors have more gastric cancer-like characteristics, while distal tumors behave “more like colorectal cancer”. How this impacts outcome and assessment of effect of therapy is not yet clear. Although analysis of our study did not reveal a statistically significant difference in survival among primary sites, it is not ruled out. Given the relative rarity of small bowel adenocarcinoma, it is unlikely that studies dedicated to site alone (e.g., duodenum) will ever be practical, even on a global scale.

The lack of celiac disease in these patients with advanced small bowel adenocarcinoma is surprising, given prior observations.²³ It could well be that celiac disease makes no significant contribution to attributable risk to small bowel adenocarcinoma, that it is as such a rare complication of a relatively rare condition. The other possibilities could be that the presence of celiac disease may permit the earlier detection of adenocarcinoma of the proximal small intestine because of gastrointestinal symptoms associated with celiac disease, thus patients are biased toward an earlier stage than those included in this study. In at least one prior observation, patients with celiac disease had an improved survival.³² Larger studies will need to be performed to identify the contribution of underlying celiac disease to SBA.

There is a strong need for collaborative studies to advance the care for rare cancers such as SBA. To that effect, the International Rare Cancers Initiative, a cooperative venture between the Cancer Research UK (CRUK), the National Cancer Institute (NCI), the European Organisation for Research and Treatment of Cancer (EORTC), the Institut National du Cancer (INCa), and the National Institute of Canada (NCI-C) has included SBA in their first round of targeted cancers.³³ Prospective trials are needed to establish standards of care for therapy of resected and metastatic SBA, and experience has shown us this can likely only be achieved through a cooperative international collaboration.

Conclusions

Dosing of irinotecan, oxaliplatin, and capecitabine in small bowel adenocarcinoma patients is tolerable, and substantial antitumor activity was noted. By dosing with a pharmacogenomic-based regimen, we demonstrated the feasibility of genotype-based dosing for a rare tumor tyoe. We observed reasonable rates of toxicity and notable antitumor activity.

Precis.

Dosing of irinotecan, oxaliplatin, and capecitabine in small bowel adenocarcinoma patients is tolerable, and substantial antitumor activity was noted. By dosing with a pharmacogenomic-based regimen, we observed reasonable rates of toxicity and signs of activity in tumor response rates.