Efficacy and safety of fruquintinib in refractory metastatic colorectal cancer: a systematic review and meta-analysis

Abstract

Background

Metastatic colorectal cancer (mCRC) remains a leading cause of cancer-related mortality, emphasizing the need for effective later-line therapies. Fruquintinib, a selective vascular endothelial growth factor receptor (VEGFR)1–3 inhibitor, has emerged as a promising option for refractory mCRC. This systematic review and meta-analysis evaluates its efficacy and safety, both as monotherapy and in combination with programmed death-1 (PD-1) inhibitors.

Methods

Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, a systematic search was conducted across PubMed, Embase, Online Vendor of International Databases (OVID), Cochrane Library, and ClinicalTrials.gov (2010–2025). Included studies were randomized controlled trials (RCTs) or real-world data on fruquintinib in mCRC after at least two prior therapies. Real-world evidence was included to complement RCT findings, as it captures broader populations, treatment patterns, and outcomes not fully reflected in controlled trial settings. Primary outcomes were progression-free survival (PFS) and overall survival (OS); secondary outcomes included objective response rate (ORR), disease control rate (DCR), and treatment-related adverse events (AEs). Pooled hazard ratios (HRs) and event rates were calculated using a random-effects model.

Results

Fifteen studies were included; 12 qualified for meta-analysis (n=3,703). Fruquintinib improved PFS [HR =0.30; 95% confidence interval (CI): 0.26–0.35] and OS (HR =0.66; 95% CI: 0.57–0.76) vs. placebo. ORR was 4.9% (95% CI: 3.2–6.6%); DCR was 62.2% (95% CI: 57.1–67.3%). Combination therapy with PD-1 inhibitors was associated with a modestly higher ORR in observational data; however, this finding requires confirmation in randomized studies (7.8% vs. 4.0%, P=0.04). In cross-study comparisons, monotherapy appeared to yield numerically longer PFS, although this was not based on head-to-head trials. AEs occurred in 86.7%, with grade ≥3 in 30.9%, most often hypertension (8.1%) and hand-foot skin reaction (5.8%). High heterogeneity was observed for several outcomes including AEs and DCR.

Conclusions

Fruquintinib significantly improves PFS and disease control in refractory mCRC with manageable toxicity. Limitations include heterogeneity across studies, with most conducted in predominantly Chinese cohorts. Further studies should explore optimal combination strategies and biomarker-based selection.

Highlight box.

Key findings

• Fruquintinib significantly improved progression-free survival [hazard ration (HR) =0.30] and overall survival (HR =0.66) in patients with refractory metastatic colorectal cancer (mCRC) compared with placebo. The pooled disease control rate was 62.2%, and the objective response rate was 4.9%. Combination therapy with programmed death-1 (PD-1) inhibitors showed a modest increase in response rates in observational data, though randomized evidence is lacking. Adverse events occurred in 86.7% of patients, with grade ≥3 events in 30.9%, most commonly hypertension and hand-foot skin reaction.

What is known and what is new?

• Fruquintinib is a selective vascular endothelial growth factor receptor (VEGFR)1–3 inhibitor approved for previously treated mCRC, with demonstrated efficacy in phase III trials, primarily in Chinese populations.

• This systematic review and meta-analysis (15 studies; n=3,703) provides the most comprehensive synthesis to date, integrating randomized and real-world data. It confirms significant survival benefit and tolerability across diverse settings, highlights modest activity when combined with PD-1 inhibitors, and underscores fruquintinib’s potential role in microsatellite-stable (MSS) disease, a population with limited immunotherapy options.

What is the implication, and what should change now?

• Fruquintinib should be considered an important third-line treatment for patients with refractory mCRC, offering survival benefit with manageable toxicity. Its emerging role in MSS tumors and combination regimens suggests the need for biomarker-driven studies and global validation. These findings support integration of fruquintinib into treatment algorithms and encourage future randomized trials of combination strategies.

Introduction

Colorectal cancer (CRC) remains a significant public health burden, ranking among the most commonly diagnosed cancers globally and the fourth leading cause of cancer-related mortality in the United States (US) (1,2). Metastatic CRC (mCRC) poses an even greater challenge, with roughly 80% of tumors unresectable (3) and a 5-year survival rate generally around 16%, as per Surveillance, Epidemiology, and End Results (SEER), despite available therapies (4). Current first- and second-line regimens rely on combinations like folinic acid-fluorouracil-oxaliplatin (FOLFOX), folinic acid-fluorouracil-irinotecan (FOLFIRI), and capecitabine-oxaliplatin (XELOX), often with endothelial growth factor receptor (EGFR) or vascular endothelial growth factor (VEGF) inhibitors (3). Guidelines from National Comprehensive Cancer Network (NCCN) and regionally Chinese Society of Clinical Oncology (CSCO) support these strategies. For patients previously treated with oxaliplatin but not irinotecan, the NCCN recommends ramucirumab plus FOLFIRI (5). Still, resistance to standard therapies persists. While third-line options, such as fruquintinib, trifluridine/tipiracil, and regorafenib, are recommended, the optimal sequencing and combination strategy remains unclear, particularly for microsatellite-stable (MSS) tumors.

Fruquintinib is a selective oral vascular endothelial growth factor receptor (VEGFR)1–3 inhibitor with minimal off-target effects on kinases such as rearranged during transfection (RET), fibroblast growth factor receptor 1 (FGFR1), and stem cell factor receptor proto-oncogene (c-kit) (6). It inhibits VEGF-driven angiogenesis to limit tumor growth. Clinical trials—including a phase Ib/II study (7) and the phase III Fruquintinib Efficacy and Safety in Colorectal Cancer (FRESCO) trial (8)—demonstrated efficacy in heavily pretreated mCRC patients, improving both overall survival (OS) and progression-free survival (PFS) compared to placebo, with manageable toxicity (8). Subgroup analyses suggest a survival benefit in patients with liver metastases (2). The FRESCO-2 trial, a global phase III study including Western populations, confirmed fruquintinib’s OS benefit, supporting its regulatory approve in the US and European Union (EU) (8,9).

Real-world studies support fruquintinib’s risk-benefit profile. One multicenter retrospective study found reduced efficacy in patients previously treated with VEGF inhibitors (5), while others confirm consistent safety and effectiveness across clinical settings (3,10). Emerging evidence supports combining fruquintinib with PD-1 inhibitors, especially in MSS and mismatch repair-proficient mCRC—subgroups that typically resist immunotherapy. Preliminary real-world and early phase data suggest potential synergistic effects when fruquintinib is combined with PD-1 inhibitors in MSS mCRC, though larger randomized trials are needed (1,6,11).

Compared to other tyrosine kinase inhibitors (TKIs) like regorafenib and apatinib, fruquintinib offers greater VEGFR selectivity, better tolerability, and improved efficacy, making it a compelling candidate for later-line therapy. Apatinib is currently approved for use in China but lacks regulatory approval for mCRC in other regions. Compared to regorafenib, a globally approved multi-kinase inhibitor, fruquintinib demonstrates greater VEGFR selectivity and a more favorable safety profile (8,12). As treatment paradigms evolve to include sequential chemotherapies, targeted therapies, immunotherapy, and supportive care, fruquintinib’s favorable profile supports integration into this continuum. This systematic review and meta-analysis evaluate its role in refractory mCRC, both as monotherapy and in combination. While limitations remain, fruquintinib represents a promising option in a field with limited alternatives. We present this article in accordance with the PRISMA reporting checklist (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-385/rc) (13).

Methods

A comprehensive search was performed across PubMed, Embase, Online Vendor of International Databases (OVID), the Cochrane Library, and ClinicalTrials.gov for studies published between January 1, 2010, and January 31, 2025. The strategy used Boolean operators and relevant medical subject headings (MeSH) and free-text terms: ((“Fruquintinib”[Title/Abstract] AND (“colorectal cancer”[MeSH Terms] OR “metastatic colorectal cancer” [Title/Abstract] OR “mCRC” [Title/Abstract]) AND (“efficacy”[Title/Abstract] OR “safety” [Title/Abstract] OR “overall survival” [Title/Abstract] OR “progression-free survival” [Title/Abstract])). Filters applied include ‘humans’, ‘Clinical Trials’ or ‘Randomized Controlled Trials’ (RCTs), ‘English’ language, publication date from January 1, 2010 to January 31, 2025. Equivalent adapted strategies were used for Embase, OVID, and the Cochrane Library.

Inclusion of real-world evidence was considered essential to complement the findings from RCTs, as it reflects broader populations, treatment patterns, and outcomes in routine clinical practice that may not be fully captured in controlled trial settings. Reference lists of included studies and related reviews were manually screened to capture additional eligible publications.

Eligibility criteria

Studies were included if they met the following predefined eligibility criteria: the study population comprised patients with refractory mCRC who received fruquintinib as a monotherapy or in combination with other agents. Primary endpoints of interest were OS and PFS. Secondary outcomes included objective response rate (ORR), disease control rate (DCR), and safety metrics based on adverse event (AE) incidence. RCTs were prioritized for inclusion in meta-analysis, given their methodological rigor in evaluating treatment efficacy and safety. However, high-quality real-world and retrospective studies were also considered and included in the systematic review and subgroup analyses, where appropriate. Exclusion criteria include: non-human studies, observation or retrospective studies, case reports, review articles, conference abstracts, studies lacking adequate outcome data, and studies published in languages other than in the English language. RCTs were prioritized for inclusion in the pooled meta-analysis, given their methodological rigor. High-quality real-world and retrospective studies were also included in the systematic review to provide complementary context. These studies were considered in narrative synthesis and designated subgroup/sensitivity analyses, but were not pooled indiscriminately with RCTs in the primary analyses.

Study selection and data extraction

Two independent reviewers conducted a blinded screening of titles and abstracts, followed by full-text review of potentially eligible studies; disagreements were resolved by consensus or adjudication by a third reviewer. The study selection process is summarized in the PRISMA flow diagram (Figure 1). Data were extracted using a standardized, pre-tested form capturing study characteristics, patient demographics, intervention details, comparators, and clinical outcomes. All data were independently cross verified by a second reviewer, with discrepancies resolved through discussion or senior investigator input.

Figure 1.

PRISMA flowchart.

Quality assessment

The methodological quality of included studies was assessed using validated risk-of-bias tools. RCTs were prioritized for quantitative synthesis, with quality evaluated using the Cochrane Risk of Bias 2.0 (RoB 2) tool, which assesses domains including selection, performance, detection, attrition, and reporting (14). High-quality real-world and retrospective cohort studies were also incorporated in predefined subgroup and sensitivity analyses to provide complementary context; these were assessed using the modified Newcastle-Ottawa Scale (NOS). Observational studies were not pooled indiscriminately with RCTs in the primary analyses but were considered separately in narrative synthesis and secondary analyses. Each study was categorized as having a low, moderate, or high risk of bias. Risk of bias ratings were based on information from published protocols, trial registries (ClinicalTrials.gov), and full-text methodology. Discrepancies were resolved by discussion or adjudicated by a third reviewer. To evaluate the potential for publication bias, both funnel plots and Egger’s test were conducted to detect small-study effects and asymmetry in outcome reporting.

Statistical analysis

For dichotomous outcomes—complete response (CR), partial response (PR), stable disease (SD), ORR, DCR, and AEs—event rates with 95% confidence intervals (CIs) were calculated using a random-effects inverse-variance model. For time-to-event outcomes such as OS and PFS, hazard ratios (HRs) with corresponding 95% CIs were determined. For studies reporting zero events, a continuity correction of 0.5 was applied prior to meta-analysis. Subgroup analyses were conducted to compare fruquintinib monotherapy with combination therapy involving PD-1 inhibitors. Sensitivity analyses were performed by sequentially removing individual studies to assess their impact on pooled estimates. Statistical significance was defined as P<0.05. Pooled estimates for event rates and HRs, along with heterogeneity statistics, were analyzed. All statistical analyses were performed using RevMan software (version 5.4.1, Cochrane Collaboration) and STATA software (version 17, StataCorp., College Station, TX, USA). The protocol for this systematic review and meta-analysis was prospectively registered in PROSPERO (ID–1059916) and conducted in accordance with PRISMA 2020 guidelines.

Results

A total of 15 studies were identified and included in the systematic review, of which 12 were eligible for inclusion in the meta-analysis, per the PRISMA guidelines. These studies collectively assessed the efficacy, safety, and survival outcomes associated with fruquintinib treatment in patients with mCRC, offering pooled data on ORR, DCR, AEs, and survival metrics.

Baseline characteristics of the studies

According to Table 1, the studies included in this analysis were predominantly conducted in China, comprising both single-center and multicenter trials with various study designs, including RCTs, prospective studies, and retrospective analyses. The sample sizes varied widely, ranging from 19 to 2,798 patients. All studies focused on patients with mCRC who had failed at least two prior chemotherapy regimens. Most patients received fruquintinib at a standard dose of 5 mg once daily for three weeks on and one week off, either as monotherapy or in combination with PD-1 inhibitors. The patient populations were generally balanced in terms of gender, with a moderate male predominance, and the median or mean age ranged from the mid-50s to mid-60s. Several studies included molecular and clinical tumor characteristics, such as rat sarcoma viral oncogene (RAS) mutation status, liver metastases, and prior surgical history, which were explored as potential predictive or prognostic factors. Of the 15 included studies, three were RCTs and the remainder were prospective or retrospective observational cohorts. Consistent with our study design, only data from RCTs were used in the primary pooled analyses of efficacy and safety outcomes, while observational cohorts contributed to subgroup and narrative analyses.

Table 1. Baseline characteristics of the studies included in the analysis.

Author, year (ref.)	Country, study design	Study population	Group	N	Male sex (%)	Age (years)#	Intervention	Tumor characteristics
								Multiple metastases (%)	Liver metastases (%)	Prior surgery (%)	RAS mutation (%)
Xu, 2017 I (7)	China, single-center, prospective	mCRC patients who had failed at least two prior chemotherapy regimens	FQ	42	59.5	55 (range, 33–70)	FQ 5 mg once daily for 3 weeks on and 1 week off	88.1	69	NR	NR
Xu, 2017 II (7)	China, single-center, RCT	mCRC patients who had failed at least two prior chemotherapy regimens	FQ	47	74.5	50 (range, 25–69)	FQ 5 mg once daily for 3 weeks on and 1 week off	95.7	61.7	NR	NR
			Placebo	24	70.8	54 (range, 38–70)	Placebo + BSC	91.7	70.8	NR	NR
Li, 2018 (8)	China, multicentric, RCT	mCRC patients who had failed at least two prior chemotherapy regimens	FQ	278	56.8	55 (range, 23–75)	FQ 5 mg once daily for 3 weeks on and 1 week off	95.3	66.5	NR	56.5
			Placebo	138	70.3	57 (range, 24–74)	Placebo + BSC	97.1	73.9	NR	53.6
Wang, 2020 (5)	China, single-center, retrospective	mCRC patients who had failed at least two prior chemotherapy regimens	FQ	46	60.9	≥60: 45.7%	At least one cycle of FQ	NR	NR	NR	54.3
Gou, 2022 (1)	China, single-center, prospective	mCRC patients who had failed at least two prior chemotherapy regimens	FQ	45	66.7	≥54: 53.3%	At least two cycles of FQ 5 mg once daily for 3 weeks on and 1 week off with PD-1 inhibitors† administered intravenously every 3 weeks	NR	80	73.3	53.3
Zhang, 2022 (6)	China, single-center, retrospective	Advanced or metastatic microsatellite-stable CRC patients who had failed at least two prior chemotherapy regimens	FQ	110	57.3	≥65: 17.3%	FQ 5 mg once daily for 3 weeks on and 1 week off with PD-1 inhibitors‡ administered intravenously every 3 weeks	54.5	80.9	NR	49.1
Dasari, 2023 (9)	International, Multicentric, Phase III RCT	mCRC patients who had failed all current standard approved cytotoxic and targeted therapies	FQ	461	53	64 (range, 56–70)	FQ 5 mg once daily for 3 weeks on and 1 week off	NR	74	NR	63
			Placebo	230	61	64 (range, 56–69)	Placebo + BSC	NR	68	NR	63
Deng, 2023 (12)	China, single-center, retrospective	mCRC patients who had failed at least two prior chemotherapy regimens	FQ	55	76.4	≥65: 45.5%	At least one cycle of FQ ± PD-1 inhibitors§	NR	70.9	NR	49.1
			RGF	50	76	≥65: 48%	At least one cycle of RGF ± PD-1 inhibitors§	NR	66	NR	44
Yang, 2023 (11)	China, single-center, retrospective	Microsatellite-stable mCRC patients who had failed at least two prior chemotherapy regimens	FQ	70	52.9	59 (range, 35–76)	FQ 5 mg once daily for 3 weeks on and 1 week off with PD-1 inhibitors¶ administered intravenously every 3 weeks	NR	72.9	74.3	52.9
He, 2025 (15)	China, single-center, retrospective	Microsatellite-stable mCRC patients with or without failure of prior chemotherapy regimens	FQ	77	54.5	62 (range, 30–80)	At least three cycles of FQ ± PD-1 inhibitors§	NR	68.8	NR	NR
Li, 2024 (10)	China, multicentric, prospective	mCRC patients who had failed at least two prior chemotherapy regimens	FQ	2,798	59.3	≥65: 35.3%	At least two cycles of FQ 5 mg once daily for 3 weeks on and 1 week off	NR	73.5	NR	NR
Wang, 2024 (16)	China, multicentric, prospective	mCRC patients	FQ	140	65.7	≥70: 24.3%	At least two cycles of FQ 5 mg once daily for 3 weeks on and 1 week off	NR	55.7	NR	65.4
Xie, 2024 I (17)	China, single-center, retrospective	mCRC patients who had failed at least two prior chemotherapy regimens	FQ	19	68.4	Mean ± SD: 58.1±9.1	FQ 5 mg once daily for 3 weeks on and 1 week off	26	42.1	NR	NR
Xie, 2024 II (17)			FQ	53	49.1	Mean ± SD: 57.0±10.4	FQ 5 mg once daily for 3 weeks on and 1 week off with PD-1 inhibitors‡	30.2	45.3	NR	NR
Xu, 2024 (18)	China, multicentric, retrospective	mCRC patients who had failed at least two prior chemotherapy regimens	FQ	520	59.8	≥65: 45.8%	At least two cycles of FQ 5 mg once daily for 3 weeks on and 1 week off with or without PD-1 inhibitors‡	87.5	71	76.2	42.5

^†, PD-1 inhibitors (200 mg pembrolizumab, 3 mg/kg nivolumab, 200 mg sintilimab, or camrelizumab); ^‡, PD-1 inhibitors (sintilimab, camrelizumab, toripalimab, tislelizumab, and pembrolizumab); ^§, PD-1 inhibitors [sintilimab (200 mg, Q3W), camrelizumab (200 mg, Q3W), tislelizumab (200 mg, Q3W), pembrolizumab (200 mg, Q3W), nivolumab (240 mg, Q3W), toripalimab (240 mg, Q3W)]; ^¶, PD-1 inhibitors (nivolumab at 240 mg every 2 weeks; pembrolizumab, tislelizumab, and sintilimab at 200 mg every 3 weeks; and toripalimab at 240 mg every 3 weeks). ^#, due to variations in the reported measures of central tendency across studies, age has been presented as originally reported in each study. BSC, best supportive care; CRC, colorectal cancer; FQ, fruquintinib; mCRC, metastatic colorectal cancer; NR, not reported; PD-1, programmed death-1; RGF, regorafenib; SD, standard deviation.

CR, PR, and SD

Eleven studies (n=1917) reported on the response to fruquintinib in mCRC. The pooled event rate of CR was 0.0% (95% CI: 0.0–0.2%; I²=0%). Per Table 2, the pooled PR rate was 4.7% (95% CI: 3.0–6.5%; I²=66.6%), with a significant difference between monotherapy and combination therapy with PD-1 inhibitors (3.7%, 95% CI: 1.2–6.1% vs. 7.8%, 95% CI: 5.1–10.4%; P=0.03) (Figure 2A). The pooled event rate of SD was 56.5% (95% CI: 51.9–61.2%; I²=70.9%). However, there was no significant difference between monotherapy and combination therapy with PD-1 inhibitors (54.6%, 95% CI: 47.4–61.7% vs. 57.0%, 95% CI: 49.7–64.4%; P=0.63).

Table 2. Outcome of interventions in the individual studies included in the analysis.

Author, year (ref.)	Group§	N	PFS (months), median (95% CI)	OS (months), median (95% CI)	CR	PR	SD	PD	ORR†	DCR‡	Comments
Xu, 2017 I (7)	FQ	42	5.8 (4.1–7.6)	8.9 (7.5–15.5)	0	4	28	7	4	32	None
Xu, 2017 II (7)	FQ	47	4.7 (2.9–5.6)	7.7 (6.9–10.3)	0	1	31	12	1	32	FQ offered significant benefit in terms of PFS over placebo
	Placebo	24	1 (0.9–1.6)	5.5 (3.6–11.3)	0	0	5	17	0	5
Li, 2018 (8)	FQ	278	3.7 (3.6–4.6)	9.3 (8.2–10.4)	1	12	160	105	13	173	FQ compared with placebo resulted in a statistically significant increase in OS
	Placebo	138	1.8 (1.8–1.8)	6.6 (5.9–8.1)	0	0	17	121	0	17
Wang, 2020 (5)	FQ	46	3.1 (1.9–4.3)	9 (7.2–10.8)	NR	NR	NR	NR	3	23	The efficacy of FQ was reduced in patients with prior use of anti-VEGFR agents
Gou, 2022 (1)	FQ	45	3.8 (2.8–4.8)	14.9 (7.6–21.7)	0	5	23	17	5	28	None
Zhang, 2022 (6)	FQ	110	5.4 (4.0–6.8)	NR	0	13	64	33	13	77	Liver metastases, ALP >160 U/L and fibrinogen >4 g/L were negative predictors of response
Dasari, 2023 (9)¶	FQ	461	3.7 (3.5–3.8)	7.4 (6.7–8.2)	0	7	249	139	7	256	None
	Placebo	230	1.8 (1.8–1.9)	4.8 (4.0–5.8)	0	0	37	143	0	37
Deng, 2023 (12)	FQ	55	4.4	14.2	0	3	29	17	3	32	FQ combined with PD-1 inhibitors showed better OS and PFS than FQ monotherapy and RGF combined with PD-1 inhibitors
	RGF (comparator group)	50	3.5	12	0	1	25	22	1	26
Yang, 2023 (11)	FQ	70	5.5 (4.3–6.8)	19.5	0	8	51	11	8	59	Liver metastasis without local treatment was a risk factor for overall survival
He, 2025 (15)	FQ	77	5.1 (3.6–6.7)	14.6 (9.6–15.6)	0	4	38	34	4	43	Prior treatment with VEGF inhibitors reduced PFS and OS
Li, 2024 (10)	FQ	2,798	NR	NR	NR	NR	NR	NR	NR	NR	None
Wang, 2024 (16)	FQ	140	6.3	12.6	0	12	57	7	12	69	Brain metastasis, sarcopenia, and baseline CEA independently predicted OS
Xie, 2024 I (17)	FQ	19	2.5 (2.2–3.8)	6.5 (4.1–8.5)	0	0	8	11	0	8	Combination therapy exhibited superior efficacy than monotherapy
Xie, 2024 II (17)	FQ	53	4.0 (2.8–8.1)	10.1 (8.0–15.2)	0	3	29	21	3	32
Xu, 2024 (18)	FQ	520	5.0 (4.3–5.8)	11.4 (10.8–22.9)	0	11	326	183	11	337	RAS mutation, BRAF mutation, and combination therapy independently predicted PFS and prior anti-VEGF and combination therapy independently predicted OS
Pooled event rates#	–	–	–	–	0.0% (0.0–0.2%)	4.7% (3.0–6.5%)	56.5% (51.9–61.2%)	30.4% (22.1–38.7%)	4.9% (3.2–6.6%)	62.2% (57.1–67.3%)	–

^†, ORR = CR + PR; ^‡, DCR = CR + PR + SD; ^§, regorafenib, included for comparison in studies with head-to-head design; ^¶, this is pivotal FRESCO-2: Global Phase III RCT in refractory mCRC, which is the basis for FDA approval; ^#, pooled event rates shown for fruquintinib-treated patients only. Comparator arms (placebo, regorafenib) were excluded from pooled AE summaries. AE, adverse event; ALP, alkaline phosphatase; CEA, carcinoembryonic antigen; CI, confidence interval; CR, complete response; DCR, disease control rate; FDA, Food and Drug Administration; FQ, fruquintinib; mCRC, metastatic colorectal cancer; NR, not reported; ORR, objective response rate; OS, overall survival; PD, progressive disease; PD-1, programmed death-1; PFS, progression-free survival; PR, partial response; RCT, randomized controlled trial; RGF, regorafenib; SD, stable disease; VEGF, vascular endothelial growth factor; VEGFR, VEGF receptor.

Figure 2.

Forest plots. (A) PR subgroup analysis; (B) ORR subgroup analysis. CI, confidence interval; DL, DerSimonian-Laird method; ORR, objective response rate; PR, partial response.

Objective response and DCR

Twelve studies (n=1,963) reported on the ORR and DCR with fruquintinib in mCRC, referring to Table 2. The pooled ORR was 4.9% (95% CI: 3.2–6.6%; I²=65.5%), with a significant difference between monotherapy and combination therapy with PD-1 inhibitors (4.0%, 95% CI: 1.6–6.4% vs. 7.8%, 95% CI: 5.1–10.4%; P=0.04) (Table 2). Analyzing data from RCTs, fruquintinib was associated with a significantly higher ORR compared to placebo with relative risk (RR) 7.53 (95% CI: 1.45–39.28; I²=0%). The pooled DCR was 62.2% (95% CI: 57.1–67.3%; I²=78.7%) (Figure 2B). However, there was no significant difference between monotherapy and combination therapy with PD-1 inhibitors (58.4%, 95% CI: 51.9–64.9% vs. 65.8%, 95% CI: 56.1–75.4%; P=0.21). Analyzing data from RCTs, fruquintinib was associated with a significantly higher DCR compared to placebo with RR 3.90 (95% CI: 3.06–4.97; I²=2%) (Figure 3).

Figure 3.

Combined subgroup analysis. The figure shows forest plots of efficacy outcomes for fruquintinib versus placebo in refractory mCRC. (A) ORR; (B) DCR; (C) OS; (D) PFS. CI, confidence interval; DCR, disease control rate; HR, hazard ratio; IV, inverse-variance method; M-H, Mantel-Haenszel method; mCRC, metastatic colorectal cancer; ORR, objective response rate; OS, overall survival; PFS, progression-free survival.

PFS and OS

Data from three RCTs (n=1,178) were used to compare the OS and PFS between fruquintinib and placebo, seen in Table 2. The forest plot showed a significant difference between both groups, with HR 0.66 (95% CI: 0.57–0.76; I²=0%) for OS and HR 0.30 (95% CI: 0.26–0.35; I²=0%) for PFS. This indicates that both OS and PFS were significantly longer with fruquintinib, suggesting survival benefits. The pooled HR for PFS with fruquintinib was 0.30 (95% CI: 0.26–0.35) compared to placebo. Similarly, the pooled HR for OS with fruquintinib was 0.66 (95% CI: 0.57–0.76) compared to placebo.

Adverse events

Referring to Table 3, ten studies (n=4,566) reported on the incidence of treatment-related AEs with fruquintinib in mCRC. The pooled incidence of AEs was 86.7% (95% CI: 80.5–92.9%; I²=98.2%) with no significant difference between monotherapy and combination therapy with PD-1 inhibitors (88.2%, 95% CI: 80.0–96.4% vs. 85.3%, 95% CI: 78.5–92.0%; P=0.59). The pooled event rate of ≥ grade 3 AEs was 30.9% (95% CI: 18.4–43.5%; I²=98.6%). The pooled incidence of fatal AEs was 2.1% (95% CI: 0.4–3.8%; I²=87.9%), with a significant difference between monotherapy and combination therapy (3.3%, 95% CI: 0.8–5.8% vs. 0.0%, 95% CI: 0.0–0.8%; P=0.01). Nine studies (n=4,047) reported on the discontinuation of treatment due to AEs. The pooled incidence of discontinuation was 16.7% (95% CI: 13.7–19.8%; I²=67.8%).

Table 3. Summary of adverse events in the individual studies included in the analysis.

Author, year (ref.)	Group	N	AE†	≥ grade 3 AE	Fatal	Hypertension (overall/≥ grade 3)	Proteinuria (overall/≥ grade 3)	HFSR (overall/≥ grade 3)	Diarrhea (overall/≥ grade 3)	Hepatotoxicity (overall/≥ grade 3)	Discontinuation
Xu, 2017 I (7)	FQ	42	42	NR	1	NR/9	NR/NR	NR/4	NR/4	NR/1	5
Xu, 2017 II (7)	FQ	47	44	NR	3	NR/14	NR/NR	NR/7	NR/1	NR/3	15
	Placebo	24	14	NR	2	NR/0	NR/NR	NR/0	NR/0	NR/2	4
Li, 2018 (8)	FQ	278	274	170	9	154/59	117/9	137/30	56/8	64/2	45
	Placebo	138	121	27	2	21/3	34/0	4/0	3/0	14/2	8
Wang, 2020 (5)	FQ	46	NR	NR	0	13/3	3/0	17/6	6/0	15/2	6
Gou, 2022 (1)	FQ	45	31	3	0	NR	NR	NR	NR	NR	NR
Zhang, 2022 (6)	FQ	110	98	15	NR	17/3	24/1	16/2	26/0	31/2	11
Dasari, 2023 (9)	FQ	461	451	286	49	168/62	79/8	88/29	110/16	48/14	93
	Placebo	230	213	116	45	20/2	12/2	6/0	24/0	11/3	49
Deng, 2023 (12)	FQ	55	43	16	0	19/14	10/3	12/4	4/0	8/0	14
	RGF	50	44	14	0	14/5	4/0	22/6	6/0	15/2	11
Yang, 2023 (11)	FQ	70	61	6	0	24/0	5/0	26/2	19/2	9/1	20
He, 2025 (15)	FQ	77	NR	NR	0	3/0	2/0	2/0	4/0	5/0	NR
Li, 2024 (10)	FQ	2,798	2,169	690	162	473/196	197/24	578/67	264/27	56/7	448
Wang, 2024 (16)	FQ	140	78	8	0	24/6	4/1	20/1	8/0	NR	15
Xie, 2024 I (17)	FQ	19	NR	NR	NR	6/NR	NR/NR	18/NR	NR/NR	NR/NR	NR
Xie, 2024 II (17)	FQ	53	NR	NR	NR	8/NR	NR/NR	27/NR	NR/NR	NR/NR	NR
Xu, 2024 I (18)	FQ	387	351	163	0	81/12	116/27	123/46	58/7	78/19	NR
Xu, 2024 II (18)	FQ	133	124	74	0	32/5	42/12	40/18	24/4	36/9	NR
Pooled event rates‡	–	–	86.7% (80.5–92.9%)	30.9% (18.4–43.5%)§	2.1% (0.4–3.8%)	25.3% (18.1–32.4%)/ 8.1% (5.1–11.1%)	16.7% (10.4–23.1%)/ 1.8% (0.8–2.9%)	32.0% (23.4–40.6%)/ 5.8% (3.7–7.9%)	14.8% (10.7–19.0%)/ 1.2% (0.5–1.9%)	17.1% (10.2–24.1%)/ 1.9% (0.8–3.0%)	16.7% (13.7–19.8%)

^†, AE denominators vary based on reporting completeness; refer to individual study protocols for specifics; ^‡, pooled event rates shown for fruquintinib-treated patients only. Comparator arms (placebo, regorafenib) were excluded from pooled AE summaries; ^§, pooled grade ≥3 AE estimates are based only on a subset of studies. AE, adverse event; FQ, fruquintinib; HFSR, hand-foot-skin reaction; N, total number of patients evaluable for AE reporting in each study; NR, not reported; RGF, regorafenib.

The pooled incidence of hand-foot-skin reaction, hypertension, hepatotoxicity, proteinuria, and diarrhea were 32.0% (95% CI: 23.4–40.6%; I²=97.4%), 25.3% (95% CI: 18.1–32.4%; I²=95.8%), 17.1% (95% CI: 10.2–24.1%; I²=96.8%), 16.7% (95% CI: 10.4–23.1%; I²=96.9%), and 14.8% (95% CI: 10.7–19.0%; I²=90.6%), respectively. The pooled incidence of ≥ grade 3 hypertension, hand-foot-skin reaction, hepatotoxicity, proteinuria, and diarrhea were 8.1% (95% CI: 5.1–11.1%; I²=93.5%), 5.8% (95% CI: 3.7–7.9%; I²=88.6%), 1.9% (95% CI: 0.8–3.0%; I²=76.3%), 1.8% (95% CI: 0.8–2.9%; I²=76.9%), and 1.2% (95% CI: 0.5–1.9%; I²=58.1%), respectively. In indirect comparisons from observational data, ≥ grade 3 hypertension was reported more frequently in monotherapy cohorts; however, these findings should be interpreted cautiously due to the lack of randomized head-to-head data (11.0%, 95% CI: 7.1–14.9% vs. 2.9%, 95% CI: 0.0–6.0%; P=0.001).

Publication bias and sensitivity analysis

Evidence of publication bias was observed for PR, ORR, and AEs, as shown in Table 4, though leave-one-out analysis did not change the overall effect size for any outcome. Positive β₁ coefficients (ORR, PR) suggest smaller studies tended to report more favorable efficacy outcomes, while negative β₁ coefficients (AEs) may indicate under-reporting of toxicity in small studies. Egger’s test results for ≥ grade 3 AEs and treatment discontinuation have wide standard errors, limiting confidence in the presence or absence of small-study effects for these outcomes. All cohort studies were rated high quality using the modified NOS, with most scoring the maximum of 7 points (Table 5), reflecting strengths in representativeness, outcome assessment, follow-up, and clarity of exposure and outcomes. An exception was Li 2024 (10), which scored lower (3/7) due to limited reporting on efficacy and follow-up. For RCTs, the RoB-2 tool indicated low risk of bias across all assessed domains—including randomization, intervention adherence, outcome data, measurement, and reporting—for Xu 2017 II (7), Li 2018 (8), and Dasari 2023 (9), supporting the methodological rigor and reliability of these trials (Table 6).

Table 4. Egger’s test for publication bias across outcomes in the meta-analysis.

Outcomes	No. of studies	β1	SE of β1	z	Prob > |z|
Complete response	13	0.05	0.375	0.14	0.88
Partial response	13	2.01	0.424	4.74	<0.001
Stable disease	13	−0.49	1.168	−0.42	0.67
Objective response rate	14	1.97	0.405	4.87	<0.001
Disease control rate	14	−1.10	1.323	−0.83	0.40
Adverse events	12	−4.15	1.703	−2.44	0.01
≥ grade 3 adverse events	10	0.32	5.654	0.06	0.95
Fatal adverse events	12	1.27	1.1298	0.98	0.32
Discontinuation	10	1.65	1.068	1.54	0.12

Evidence of publication bias for the outcomes of partial response, objective response rate, and adverse events. A P value <0.05 was considered statistically significant for Egger’s test. Egger’s test may be underpowered and less reliable for outcomes with fewer than 10 studies. β₁: regression coefficient from Egger’s test, indicating the degree of funnel plot asymmetry (potential publication bias). SE: standard error of the regression coefficient (β₁), reflecting variability of the estimate. z: test statistic calculated as β₁ divided by its standard error. Prob > |z|: P value corresponding to the z statistic, testing the null hypothesis of no small-study effects (no publication bias). |z|: absolute value of the z statistic, reported for ease of interpretation regardless of direction.

Table 5. Study quality assessment using modified NOS for cohort studies.

Author, year (ref.)	Representative of the exposed cohort	Ascertainment of exposure	Demonstration that outcome of interest was not present at start of study	Definite information on efficacy and safety	Assessment of outcome	Was follow-up long enough for outcomes to occur	Adequacy of follow-up	Category [total score]†
Xu, 2017 (7)	*	*	*	*	*	*	*	High [7]
Wang, 2020 (5)	*	*	*		*			Medium [4]
Gou, 2022 (1)	*	*	*		*	*	*	High [6]
Zhang, 2022 (6)	*	*	*	*	*	*		High [6]
Deng, 2023 (12)	*	*	*	*	*			Medium [5]
Yang, 2023 (11)	*	*	*	*	*	*		High [6]
He, 2025 (15)	*	*	*	*	*	*		High [6]
Li, 2024 (10)	*	*	*		*	*	*	High [6]
Wang, 2024 (15)		*	*	*	*	*	*	High [6]
Xie, 2024 (17)	*	*	*		*	*		Medium [5]
Xu, 2024 (18)	*	*	*	*	*	*		High [6]

The table demonstrates that all cohort studies included in the analysis were rated as high quality using the modified NOS, with most studies scoring the maximum of 7 points. Of note, “*” indicates that the study met the criterion and was awarded a point on the modified NOS. The total score is the sum of awarded points across all categories (maximum =7). Two independent reviewers assessed study quality using the modified NOS. Discrepancies were resolved by consensus or adjudicated by a third reviewer. Scoring was based on the modified NOS for cohort studies: Wells et al. (19). ^†, maximum =7; high >5; medium 4–5; low <4. NOS, Newcastle-Ottawa Scale.

Table 6. Study quality assessment using ROB-2 for randomized controlled.

Author, year (ref.)	Randomization process	Deviations from intended interventions	Missing outcome data	Measurement of the outcome	Selection of the reported result	Overall
Xu, 2017 II (7)	Low risk	Low risk	Low risk	Low risk	Low risk	Low risk
Li, 2018 (8)	Low risk	Low risk	Low risk	Low risk	Low risk	Low risk
Dasari, 2023 (9)	Low risk	Low risk	Low risk	Low risk	Low risk	Low risk

Per the table, all RCTs, using Cochrane Risk of Bias 2.0 (ROB-2) tool, indicated low risk of bias across all evaluated domains. No outcome switching was observed when comparing published outcomes to pre-registered trial protocols. RCTs, randomized controlled trials.

Discussion

Fruquintinib shows encouraging efficacy signals as a third-line option for refractory mCRC, offering survival benefits, disease control, and immunotherapy compatibility. This section outlines its efficacy, safety, and clinical impact.

Fruquintinib as a therapeutic option

Fruquintinib is a selective TKIs targeting VEGFR1, VEGFR2, and VEGFR3, approved for mCRC patients who progressed after at least two prior systemic therapies. It received approval in China in 2018 and the US in 2020. Administered orally at 5 mg daily, fruquintinib inhibits VEGF-mediated angiogenesis, curbing tumor neovascularization and metastatic spread (7,16,17). Its approval was based on the FRESCO trial, which showed significant PFS and OS improvements in the third-line setting. Multiple clinical and real-world studies have shown better efficacy of fruquintinib’s efficacy across OS, PFS, and DCR, reinforcing its role in late line mCRC therapy (6).

Efficacy of fruquintinib monotherapy

Fruquintinib’s monotherapy efficacy has been demonstrated through clinical trials that are further elaborated in the discussion. Xu et al., in phase Ib of their RCT, reported a median PFS of 5.80 months and OS of 8.88 months (7). A subsequent phase II trial confirmed benefit, with median PFS of 4.73 vs. 0.99 months (HR =0.71) and median OS of 7.72 vs. 5.5 months (7).

The phase III FRESCO trial, a multicenter, double-blind study, reinforced this, showing OS of 9.3 vs. 6.6 months (HR =0.65) and PFS of 3.7 vs. 1.8 months (HR =0.26) (8). A subgroup analysis of patients with colorectal liver metastases (CRLM) showed OS of 8.61 vs. 5.98 months, highlighting fruquintinib’s relevance in this population (5).

Real-world data by Jin et al., comparing FRESCO findings with other TKIs, found a median PFS of 3.71 months for fruquintinib vs. 2.49 months (3), with enhanced benefit in left-sided colon cancer, multiple metastases, or lung involvement. These results support fruquintinib’s efficacy in the third-line setting, particularly in specific mCRC subgroups.

Standard fruquintinib regimen was consistently reported across included studies. However, no dose-response analyses were performed in our review, and therefore the potential influence of dose intensity or modifications on efficacy and safety outcomes could not be assessed. This represents a limitation in interpreting the findings.

Fruquintinib in combination therapy

Emerging evidence supports combining fruquintinib with PD-1 inhibitors. The rationale lies in synergistic mechanisms—fruquintinib remodels the tumor microenvironment to enhance T-cell infiltration and tumor immunogenicity, boosting the efficacy of immune checkpoint blockade (11,15).

Real-world data suggests this strategy may be beneficial. Gou et al. reported an ORR of 11.1%, DCR of 62.2%, median PFS of 3.8 months, and OS of 14.9 months with fruquintinib plus PD-1 (1). In their retrospective study, Xu et al. found even stronger results: OS of 19.8 months, PFS of 5.5 months, and DCR of 84.3% in MSS mCRC (18).

In a comparative study, Deng et al. showed that fruquintinib plus PD-1 outperformed regorafenib plus PD-1 in ORR (6.1% vs. 2.0%) and DCR (65.3% vs. 54.2%), with improved OS and PFS (12,20). Fruquintinib’s high VEGFR2/3 selectivity, low 5 mg dosing, and tolerability contribute to its superior therapeutic index and make it more suitable for long-term use (12,15,18,20).

Fruquintinib in MCRC

Liver and lung metastases in mCRC are associated with worse prognosis. Local interventions, when combined with VEGFR inhibitors like fruquintinib, may improve survival. Yang et al. showed that patients undergoing liver-directed therapy plus VEGFR inhibitors had better OS; lack of local treatment was linked to worse survival (HR =5.31) (11). Patients receiving local liver treatment reached OS comparable to those without hepatic metastases, possibly due to enhanced hepatic function and systemic immunity (11). Zhang et al. corroborated these findings with an ORR of 11.8%, a DCR of 70%, and a median PFS of 5.4 months. Adverse prognostic markers included elevated alkaline phosphatase (ALP) (>160 U/L, HR =0.478), fibrinogen (FIB) (≥4 g/L, HR =0.517), and rising ALP during treatment (HR =1.673), along with liver metastases (HR =0.594) (6).

Importantly, prior VEGFR inhibitor exposure reduced fruquintinib OS benefit but did not impact its efficacy in shrinking liver metastases (21). Biomarkers such as FIB, FIB degradation product (D-dimer), and ALP may help predict fruquintinib response, particularly in combination strategies. Their low cost and ease of measurement make them promising tools for guiding therapy (6,22).

Fruquintinib also shows efficacy in MSS mCRC, a subgroup unresponsive to immune checkpoint inhibitors. While most studies in this review did not stratify by microsatellite status, those that did reported benefit in MSS patients. Fruquintinib’s VEGFR-driven mechanism explains this efficacy, independent of immunogenic pathways. Combination regimens with PD-1 inhibitors may enhance outcomes in MSS tumors, but further biomarker-driven studies are needed to clarify its role. These results highlight an unmet need in MSS mCRC, where fruquintinib may offer new options for patients with limited alternatives. Most studies included in this review did not stratify results by microsatellite instability status. Given that the majority of mCRC cases are MSS, and treatment options for this subgroup remain limited, the role of fruquintinib in MSS disease warrants further prospective evaluation.

Adverse effects and safety profile

Fruquintinib’s safety profile supports its third-line use. A phase IV real-world study found treatment-emergent AEs in 76.2% of patients, with 23.9% experiencing grade ≥2 treatment related AEs. Grade 3 events included hypertension (6.6%), palmar erythrodysesthesia (2.2%), and thrombocytopenia (1.0%). These were the main reasons for dose reductions or interruptions. Other side effects included anorexia, diarrhea, liver dysfunction, and anemia—generally manageable with supportive care (6). In indirect comparisons from observational data, ≥ grade 3 hypertension was reported more frequently in monotherapy cohorts; however, these findings should be interpreted cautiously due to the lack of randomized head-to-head data.

Compared to other VEGFR inhibitors, fruquintinib has a lower treatment related AEs incidence and treatment-related mortality (0.3%) (6). Rare toxicities like hypothyroidism and immune-mediated pneumonitis have been reported but are infrequent and manageable (6).

A comparative study found that fruquintinib had a similar overall toxicity profile to regorafenib. While hypertension was more common with fruquintinib, hand-foot syndrome was more frequent with regorafenib. When used in combination with PD-1 inhibitors, fruquintinib did not significantly increase toxicity, supporting its use in such regimens (5). Overall, fruquintinib is well tolerated, with reversible and predictable AEs. This strengthens its candidacy as a third-line therapy in mCRC.

Interpretation of all the pooled results is limited by substantial heterogeneity across studies in terms of design, treatment context, patient populations, and prior lines of therapy. Assessment of publication bias using funnel plots (Figure S1) revealed asymmetry for certain efficacy and safety outcomes, consistent with Egger’s test findings, suggesting that smaller studies may have overestimated treatment effects. Future research should aim for harmonized outcome reporting and stratified analyses to enhance comparability.

Conclusions

Fruquintinib shows promise as a third-line option in mCRC, with early evidence supporting efficacy and tolerability, especially in patients who have exhausted standard therapies. However, further randomized studies are needed to confirm these findings across a broader and more diverse patient population. Our systematic review and meta-analysis highlight its clinical benefit, demonstrating notable improvements in OS and PFS compared to placebo, along with a favorable safety profile. Beyond monotherapy, early combination strategies involving fruquintinib and PD-1 inhibitors show potential in observational cohorts; however, these findings require confirmation through prospective, controlled trials.

Preliminary real-world comparisons suggest that fruquintinib may offer improved selectivity and tolerability compared to regorafenib, though direct comparative trials are needed to confirm this. However, limitations include heterogeneity in study design, predominance of data from Chinese populations, and lack of long-term follow-up. Future research should focus on biomarker-driven approaches, optimal combination strategies, and validating efficacy across a diverse global population. In light of the lack of effective immunotherapy options in MSS mCRC, fruquintinib, particularly in combination with PD-1 inhibitors, represents a potential strategy that merits further biomarker-guided investigation. Another limitation of our analysis is the lack of randomized evidence for combination therapy and the heterogeneity across study designs restrict the strength and generalizability of these findings. Interpretation of our pooled estimates is limited by substantial heterogeneity (with I² values often exceeding 60–90% across outcomes such as ORR, DCR, and AEs) and by the predominance of Chinese cohorts (>90% of included patients), which may restrict global generalizability. Although random-effects, leave-one-out, and sensitivity analyses confirmed the overall direction of effect, these findings should be interpreted with caution, underscoring the need for harmonized reporting and validation in more diverse populations. Given ongoing challenges in refractory mCRC, additional studies are needed to optimize its use—particularly in combination regimens and biomarker-defined populations. As treatment paradigms evolve, fruquintinib holds strong potential to improve survival and quality of life in this difficult-to-treat setting.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.