Why This Study Matters
Chemotherapy-induced thrombocytopenia (CIT) is a common, challenging complication of cytotoxic chemotherapy. The authors note there are no widely available, licensed agents for its management, so reducing chemotherapy relative dose intensity through delays, dose reductions, and agent discontinuation remains the standard of care. Persistent CIT, defined here as a platelet count ≤85 × 10⁹/L on day 1 of a scheduled chemotherapy cycle despite adequate time to recover from the prior cycle, is particularly difficult to manage. Avatrombopag is a once-daily oral second-generation thrombopoietin receptor agonist (TPO-RA).
47 patients enrolled; 4 US trial sites; 1:1 randomization; 40 mg starting daily dose of avatrombopag.
Study Design
ACT-GI (NCT05772546) was a multicenter, randomized, double-blind, placebo-controlled, investigator-initiated phase II trial at four US sites. Patients were randomly assigned 1:1 to avatrombopag or matched placebo, stratified by number of cytotoxic agents (1 vs ≥2). After a lead-in period of up to 14 days starting at 40 mg once daily, patients who recovered to a platelet count of ≥100 × 10⁹/L resumed chemotherapy at the same dosing for one on-study cycle. Doses were titrated weekly between a minimum of 20 mg three times per week and a maximum of 60 mg once daily, targeting 100 to 250 × 10⁹/L. Enrollment ran October 4, 2023 to October 10, 2025; data cutoff was January 2, 2026. The design targeted 60 participants (30 per group).
Funding: Sobi provided a grant to Massachusetts General Hospital and supplied avatrombopag and matching placebo; the trial was investigator-initiated.
Patient Population
Adults receiving cytotoxic chemotherapy for gastroesophageal, small bowel, colorectal, pancreatic, or hepatobiliary cancers (excluding GI lymphomas). Forty-seven patients were enrolled (23 avatrombopag, 24 placebo). The trial was closed by the data and safety monitoring board at the planned interim analysis after meeting the prespecified stopping criterion for efficacy (Z = 3.87, P < .001).
| Baseline | Placebo (n = 24) | Avatrombopag (n = 23) |
|---|---|---|
| Female, No. (%) | 10 (42) | 4 (17) |
| Age, median (range), years | 59 (39-77) | 63 (25-84) |
| Colorectal, No. (%) | 16 (67) | 13 (57) |
| Stage 4, No. (%) | 21 (88) | 17 (74) |
| Chemotherapy dose reduction before enrollment, No. (%) | 12 (50) | 15 (65) |
| Qualifying platelet count, median (range), ×10⁹/L | 70 (44-83) | 69 (22-83) |
Primary Endpoint
The primary endpoint was successful CIT treatment and prevention of recurrence: reaching a platelet count ≥100 × 10⁹/L within the 2-week lead-in, completing the on-study chemotherapy cycle with no thrombocytopenia-related dose reduction or delay, and reaching ≥100 × 10⁹/L at the start of the following cycle. Platelet transfusion at any time counted as treatment failure. The endpoint was met by 16 of 23 patients (70% [95% CI, 47 to 87]) on avatrombopag and 4 of 24 (17% [95% CI, 5 to 37]) on placebo (Z = 3.67, P < .001). Median (IQR) platelet count at the end of the on-cycle period was 157 (136-202) × 10⁹/L with avatrombopag versus 72 (68-134) × 10⁹/L with placebo.
Subgroup and Sensitivity Analyses
The paper reports post hoc sensitivity analyses of the primary endpoint (Appendix Table A4) and describes findings as similar across chemotherapy backbone and cycle length. No formal subgroup hypothesis tests were reported.
| Analysis (post hoc) | Avatrombopag responders | Placebo responders | P |
|---|---|---|---|
| 14-day cycle | 13/18 (72%) | 3/18 (17%) | <.001 |
| 21- or 28-day cycle | 3/5 (60%) | 1/6 (17%) | NA |
| Oxaliplatin-based | 8/12 (67%) | 4/18 (22%) | — |
| Not oxaliplatin-based | 8/11 (73%) | 0/6 (0%) | — |
Within-arm comparison of oxaliplatin-based versus not: avatrombopag P > .9; placebo P = .5. Numbers in these strata are very small.
Safety Profile
| Adverse event | Placebo (n = 24) | Avatrombopag (n = 23) |
|---|---|---|
| Any AE | 11 (46%) | 18 (78%) |
| Treatment-related AE | 2 (8%) | 3 (13%) |
| Grade ≥3 | 2 (8%) | 5 (22%) |
| Serious AE | 0 (0%) | 4 (17%) |
| Leading to study drug discontinuation | 0 (0%) | 1 (4%) |
| Headache | 2 (8%) | 4 (17%) |
| Diarrhea | 0 (0%) | 3 (13%) |
| Nausea | 0 (0%) | 3 (13%) |
| Neutrophil count decreased | 3 (13%) | 0 (0%) |
No grade ≥3 or serious AEs were study drug-related, and no AEs of thrombocytosis were reported. One patient in the avatrombopag group, with prior cerebrovascular disease and ischemic strokes, had a fatal ischemic stroke approximately 30 minutes after the first dose, assessed as unrelated to study drug. There were no venous thromboembolic events. One patient in the avatrombopag group had a clinically relevant bleeding event (grade 2 intestinal stoma site bleed) unrelated to thrombocytopenia.
Interpretation and Broader Context
The authors conclude that the study demonstrated efficacy of avatrombopag in persistent CIT, noting that a prior placebo-controlled trial of avatrombopag in 122 patients with mostly mid-cycle nadir CIT (AVA-CIT-330) failed its primary endpoint because of high spontaneous platelet recovery in the placebo group.
In conclusion, avatrombopag was safe and efficacious in the management of persistent CIT in patients with GI cancers.
— Conclusion of Al-Samkari et al., J Clin Oncol 2026
Limitations
The authors state that the study did not evaluate survival outcomes, that bleeding and thromboembolic event rates are too low for meaningful comparison, and that it remains unproven whether dose-intensity gains from correcting CIT improve survival. The open-label extension is ongoing. The trial enrolled 47 patients, short of its target of 60, because it was stopped at the planned interim analysis for efficacy. Baseline imbalances (for example, 42% vs 17% female) are visible in Table 1 of a small trial. The subgroup and sensitivity analyses were post hoc, with very small numbers per stratum. This article does not make treatment recommendations.