Robotic-assisted donor and recipient hepatectomy in liver transplantation: An umbrella review of clinical outcomes, surgical performance, and cost-effectiveness

Table 1 Characteristics of included reviews

Ref.	Country	Population (scope)	Intervention	Comparator	Outcomes	Type of synthesis
Hobeika et al[20], 2025	France	HPB robotic surgery (liver & pancreas); transplantrelevant questions	Robotic hepatectomy	Open, laparoscopic	Practice recommendations; clinical safety/feasibility	Consensus guideline with embedded systematic review (context only; not graded)
Giglio et al[21], 2025	Italy	Living liver donors (RADH)	Robotic donor hepatectomy	Open ± laparoscopic	Clinical outcomes; surgical performance	Systematic review + meta-analysis
Pilz da Cunha et al[22], 2025	Netherlands	Adults undergoing liver resection (nontransplant); used as indirect evidence for recipients	RLR	Laparoscopic (LLR) ± open	Clinical outcomes (conversion, morbidity, LOS, etc.)	Systematic review + meta-analysis
Koh et al[23], 2024	Singapore	Adults undergoing liver resection (nontransplant); used as indirect evidence for recipients	Robotic (RLR) (plus LLR/OLR in network)	Network across RLR, LLR, OLR	Clinical outcomes; costs (total, procedural, hospitalization); costeffectiveness	Network meta-analysis (clinical + economic)
Broering et al[24], 2024	Germany	Minimally invasive organ transplantation (multiorgan); includes liver donor/recipient subset	Robotic and laparoscopic approaches (donor/recipient)	Open and/or laparoscopic	Clinical, surgical performance, cost (pooled where available)	Systematic review + meta-analysis (perspective with embedded systematic review + meta-analysis)

RLR: Robotic liver resection; LLR: Laparoscopic resections; OLR: Open surgery; HPB: Hepato-pancreato-biliary; RADH: Robotic-assisted donor hepatectomy; LOS: Length of stay.

Table 2 Summary of findings - living liver donors (robotic vs open/Laparoscopic donor hepatectomy; direct evidence)

Outcome	Assumed risk (comparator per 1000)	Corresponding risk (robotic per 1000)	Effect [RR/OR/MD (95%CI)]	Number of studies (participants)	Certainty (GRADE)	Notes
Overall complications	Not estimable (baseline risk not reported)	Not estimable (requires baseline or pooled RD)	RR = 0.85 (0.72-0.99)	6 studies; 2168 donors	Low	1,2,4
Major complications (Clavien-Dindo ≥ III)	Not estimable (baseline risk not reported)	Not estimable (requires baseline or pooled RD)	RR = 0.80 (0.65-0.98)	6 studies; 2168 donors	Low	1,2,4
Minor complications (Clavien-Dindo I)	Not estimable (baseline risk not reported)	Not estimable (requires baseline or pooled RD)	OR = 5.14 (1.06-24.93) (open vs robotic)	2 studies; 216 donors	Low	1,3,4
Operative time (minutes)	N/A (continuous outcome)	N/A (continuous outcome)	MD = +105 (68142) minutes	6 studies; 2168 donors	Low	1,2
Blood loss (mL)	N/A (continuous outcome)	N/A (continuous outcome)	MD = -117 (-229 to -6) mL	6 studies; 2168 donors	Low	1,2
Length of stay (days)	N/A (continuous outcome)	N/A (continuous outcome)	MD = -0.60 (-1.14 to -0.10) days	6 studies; 2168 donors	Low	1,2

¹Risk of bias.

²Inconsistency/heterogeneity.

³Imprecision.

⁴Absolute effect not estimable from ratio-only data without a pooled baseline (or when risk difference was not reported).

Continuous outcomes use mean differences; absolute-risk columns are not applicable. Grading of recommendations assessment, development and evaluation downgrades reflect observational designs and heterogeneity in pooled estimates. Heterogeneity (Giglio et al[21]): Blood loss [open vs robotic weighted mean difference (WMD) 117.54 mL → -117.54 mL for robotic vs open]: I² = 98%; τ² = 18847.102. Operative time (open vs robotic WMD -105.34 minutes → +105.34 minutes): I² = 91%; τ² = 1772.424. Length of stay (open vs robotic WMD 0.63 days → -0.63 days): I² = 82%; τ² = 0.259. Donor complications were reported only as ratio measures without pooled baseline risks; absolute effects are not estimable. RR: Risk ratios; MD: Mean difference; OR: Odds ratio; RD: Risk difference; N/A: Not applicable; GRADE: Grading of recommendations assessment, development and evaluation.

Table 3 Recipients and economics (indirect evidence) - summary of findings and certainty (grading of recommendations assessment, development and evaluation): Recipients (indirect)

Outcome	Assumed risk (comparator per 1000)	Corresponding risk (robotic per 1000)	Effect, RR/MD; RD when available (95%CI)	Number of studies (participants)	Certainty (GRADE)	Notes
Conversion to open	RD pooled (no baseline required)	-50 per 1000	RR = 0.41 (0.32-0.52); RD = -0.05 (-0.07 to -0.03)	24 studies; participants NR	Low-moderate	1,3
Overall morbidity (any)	RD pooled (no baseline required)	-20 per 1000	RR = 0.92 (0.84-1.00); RD = -0.02 (-0.03 to 0.00)	20 studies; participants NR	Low	1,4
Severe morbidity (Clavien-Dindo ≥ III)	RD pooled (no baseline required)	-10 per 1000	RR = 0.81 (0.70, 0.94); RD ≈ -0.01	21 studies; participants NR	Low-moderate	1,3
Readmission	RD pooled (no baseline required)	+10 per 1000	RR = 1.24 (1.09, 1.41); RD = +0.01 (0.00, 0.01)	20 studies; participants NR	Low	1,2
30day mortality	Not estimable	Not estimable	RR = 0.77 (0.51, 1.15)	20 studies; participants NR	Low	1,4,5
Operative time (minutes)	N/A	N/A	MD = +8.8 (-4.8, 22) minutes	23 studies; participants NR	Low	1,2
Length of stay (days)	N/A	N/A	MD = -0.27 (-0.67, 0.12) days	26 studies; participants NR	Low	1,2

¹Indirectness: Recipient outcomes and all economic contrasts derive from non-transplant liver resections; downgraded for indirectness in grading of recommendations assessment, development and evaluation.

²Baseline risk: Absolute effects per 1000 are reported only when pooled baseline/comparator risks or pooled risk difference were available; otherwise reported as risk difference or “Not estimable”.

³Economics: Costs are taken verbatim from the network meta-analysis without re-standardizing baskets; interpret as hypothesis-generating for transplant settings.

⁴Context (not graded): MIOT meta-analysis and consensus guideline inform field context only and are not included in certainty ratings.

⁵Imprecision/inconsistency where applicable (e.g., mortality).

Table 3 (indirect): Robotic liver resection (RLR) vs laparoscopic/open. Heterogeneity: Extracted from Pilz da Cunha et al[22] and Koh et al[23]. Recipients (robotic vs laparoscopic resections): Overall morbidity I² = 24%; severe morbidity I² = 7%; 30-day mortality I² = 0%; operative time I² = 90%; length of stay I² = 41% (τ² not reported in the article text). Conversion: Number needed to treat ≈ 20; I² not reported in the article text/tables (likely in Supplementary material). Economics: The network meta-analysis reports comparative costs but does not provide standard I²/τ² for cost nodes; heterogeneity not estimable across cost categories. RR: Risk ratios; MD: Mean difference; GRADE: Grading of recommendations assessment, development and evaluation; RD: Risk difference; N/A: Not applicable; NR: Not reported.

Table 4 Recipients and economics (indirect evidence) - summary of findings and certainty (grading of recommendations assessment, development and evaluation): Economics (indirect)

Economic outcome	Contrast	Effect, MD (95%CI/CrI)	Source	Certainty (GRADE)	Notes
Total costs	RLR vs OLR	-243.82 (-326.46 to -162.78)	Network meta-analysis (Koh et al[23], 2024)	Low-moderate	1,2
Total costs	LLR vs OLR	-217.48 (-282.44 to -154.94)	Network meta-analysis (Koh et al[23], 2024)	Low-moderate	1,2
Procedure costs	RLR vs LLR	+2310.17 (+812.98 to +3808.74)	Network meta-analysis (Koh et al[23], 2024)	Low-moderate	1,2
Hospitalization costs	RLR vs OLR	-3527.20 (-5760.20 to -1367.50)	Network meta-analysis (Koh et al[23], 2024)	Low-moderate	1,2

¹Indirectness: Recipient outcomes and all economic contrasts derive from non-transplant liver resections; downgraded for indirectness in grading of recommendations assessment, development and evaluation.

²Economics: Costs are taken verbatim from the network meta-analysis without re-standardizing baskets; interpret as hypothesis-generating for transplant settings.

Table 4 (indirect): Economic evidence from non-transplant liver resections; hypothesis-generating for transplant. Absolute effects for dichotomous outcomes are shown when pooled baseline/comparator risks or pooled risk difference were available; otherwise “not estimable”. Heterogeneity: Extracted from Pilz da Cunha et al[22] and Koh et al[23]. Recipients (robotic vs laparoscopic resections): Overall morbidity I² = 24%; severe morbidity I² = 7%; 30-day mortality I² = 0%; operative time I² = 90%; length of stay I² = 41% (τ² not reported in the article text). Conversion: Number needed to treat ≈ 20; I² not reported in the article text/tables (likely in Supplementary material). Economics: The network meta-analysis reports comparative costs but does not provide standard I²/τ² for cost nodes; heterogeneity not estimable across cost categories. MD: Mean difference; GRADE: Grading of recommendations assessment, development and evaluation; USD: United States Dollars; RLR: Robotic liver resection; LLR: Laparoscopic resections; OLR: Open surgery.

Table 5 Methodological quality of included reviews (AMSTAR 2) - summary: AMSTAR 2 overall confidence for each review

Ref.	Overall AMSTAR 2 rating	Key strengths/Limitations
Hobeika et al[20]	Moderate	PROSPERO-registered SR; multi-database search; dual/blinded quality appraisal; no de novo quantitative pooling for transplant-specific outcomes
Giglio et al[21]	Moderate	Protocol/registration; duplicate screening and extraction; sensitivity analyses; limited grey literature reporting
Pilz da Cunha et al[22]	High	Comprehensive search; duplicate processes; risk-of-bias assessment; appropriate meta-analytic methods; sensitivity/subgroup analyses
Koh et al[23]	High	Transparent network meta-analysis; study quality and model diagnostics described; heterogeneity and small-study effects assessed
Broering et al[24]	High	Explicit multi-database search with PRISMA flow; pooled estimates with heterogeneity; consideration of study quality; publication-bias assessment where feasible

Item-level checklists and justifications are provided in Supplementary Table 3. For Hobeika et al[20], AMSTAR 2 pertains exclusively to the embedded systematic-review methods; the consensus recommendations were not graded.

Table 6 Summary of findings (grading of recommendations assessment, development and evaluation) - robotic-assisted hepatectomy vs comparators in liver transplantation: Living liver donors - robotic vs open donor hepatectomy

Outcome	Assumed risk (comparator per 1000)	Corresponding risk (robotic per 1000)	Effect [RR/MD (95%CI)]	No. of studies (participants)	Certainty (GRADE)	Comments/footnotes
Operative time (minutes)	N/A (continuous outcome)	N/A (continuous outcome)	MD = +105 (68142)	6 studies; 2168 donors	Low	Robotic longer operative time; source Giglio et al[21], 2025 (ODH vs RADH section)
Blood loss (mL)	N/A (continuous outcome)	N/A (continuous outcome)	MD = -117 (-229 to -6)	6 studies; 2168 donors	Low	Robotic less blood loss; Giglio et al[21], 2025
Length of stay (days)	N/A (continuous outcome)	N/A (continuous outcome)	MD = -0.60 (-1.14 to -0.10)	6 studies; 2168 donors	Low	Robotic shorter LOS; Giglio et al[21], 2025
Minor complications (Clavien-Dindo I)	Not estimable (baseline risk not reported)	Not estimable (requires baseline or pooled RD)	OR = 5.14 (1.06-24.93)	2 studies; 216 donors	Low	OR 5.14 (1.06-24.93) (open vs robotic). Baseline not reported; absolute effect not estimable

N/A: Not applicable; GRADE: Grading of recommendations assessment, development and evaluation; ODH: One digital health; RADH: R-specific alcohol dehydrogenase; LOS: Length of stay; RR: Risk ratios; MD: Mean differences; RD: Risk difference.

Table 7 Summary of findings (grading of recommendations assessment, development and evaluation) - robotic-assisted hepatectomy vs comparators in liver transplantation: Transplant recipients - robotic vs laparoscopic

Outcome	Assumed risk (comparator per 1000)	Corresponding risk (robotic per 1000)	Effect [RR/MD (95%CI)]	Number of studies (participants)	Certainty (GRADE)	Comments/footnotes
Conversion to open	RD pooled (no baseline required)	-50 per 1000	RR 0.41 (0.32-0.52); RD -0.05 (-0.07 to -0.03)	24 studies; participants NR	Low-moderate	Pilz da Cunha et al[22], 2025; NNT = 20
Overall morbidity (any complication)	RD pooled (no baseline required)	-20 per 1000	RR 0.92 (0.84-1.00); RD -0.02 (-0.03 to 0.00)	20 studies; participants NR	Low	Pilz da Cunha et al[22], 2025 (overall morbidity section)
Severe morbidity (Clavien-Dindo ≥ III)	RD pooled (no baseline required)	-10 per 1000	RR 0.81 (0.70, 0.94); RD ≈ -0.01 (NNT = 100)	21 studies; participants NR	Low-moderate	Pilz da Cunha et al[22], 2025; NNT = 100 overall; subgroup major RD -0.03
R0 resection rate	Not estimable (RD not provided)	Not estimable	RR 1.02 (1.01-1.03)	NR (multi-study)	Low	Pilz da Cunha et al[22], 2025 (counts not explicit in text extraction)
Readmission	RD pooled (no baseline required)	+10 per 1000	RR 1.24 (1.09-1.41); RD +0.01 (0.00-0.01)	20 studies; participants NR	Low	Pilz da Cunha et al[22], 2025
Length of stay (days)	N/A (continuous outcome)	N/A (continuous outcome)	MD -0.27 (-0.67 to 0.12)	26 studies; participants NR	Low	Pilz da Cunha et al[22], 2025
30-day or in-hospital mortality	Not estimable (RD not provided)	Not estimable (requires baseline risk)	RR 0.77 (0.51-1.15)	20 studies; participants NR	Low	Pilz da Cunha et al[22], 2025 (30-day mortality)
Blood loss (mL)	N/A (continuous outcome)	N/A (continuous outcome)	No significant difference (overall)	NR	Low	Pilz da Cunha et al[22], 2025 (overall non-significant; subgroup differences exist)
Operative time (minutes)	N/A (continuous outcome)	N/A (continuous outcome)	MD +8.8 (-4.8 to 22)	23 studies; participants NR	Low	Pilz da Cunha et al[22], 2025; high heterogeneity (I2 ≈ 90%)

NR: Not reported; N/A: Not applicable; NNT: Number needed to treat; GRADE: Grading of recommendations assessment, development and evaluation; RR: Risk ratios; MD: Mean differences; RD: Risk difference.

Table 8 Summary of findings (grading of recommendations assessment, development and evaluation) - robotic-assisted hepatectomy vs comparators in liver transplantation: Economic/resource outcomes - robotic vs comparators (as reported in Koh et al[23], 2024)

Outcome	Effect estimate (MD/OR/NMA contrast)	Certainty (GRADE)	Comments/footnotes
Total costs	RLR vs OLR: MD -243.82 (-326.46 to -162.78)	Low-moderate	Less intraoperative blood loss with RLR vs OLR also reported
Total costs (LLR vs OLR)	MD -217.48 (-282.44 to -154.94)	Low-moderate	Koh et al[23], 2024 network meta-analysis
Procedure costs	RLR vs LLR: MD + 2310.17 (+812.98 to +3808.74)	Low-moderate	Downgrades: RoB (observational inputs); indirectness to transplant setting; currency/modeling heterogeneity1,2
Hospitalization costs	RLR vs OLR: MD -3527.20 (-5760.20 to -1367.50)	Low-moderate	Lower hospitalization costs for RLR vs OLR
Transfusion	RLR vs OLR: OR 0.34 (0.17-0.66)	Low-moderate	Koh et al[23], 2024 NMA
Overall morbidity	RLR vs OLR: OR 0.41 (0.26-0.59)	Low-moderate	Lower morbidity vs OLR

¹Risk of bias (observational comparative designs).

²Indirectness (non-transplant or mixed populations).

Grading of recommendations assessment, development and evaluation-compliant summary-of-findings table. Relative and absolute effects are reported where available. Continuous outcomes are presented as mean differences. For dichotomous outcomes, absolute risks per 1000 will be derived from the median comparator risk of contributing meta-analyses; cells are footnoted when baseline risk is pending. Donor meta-analytic lines from Giglio et al[21], 2025 (open vs robotic donor hepatectomy). Recipient effects from Pilz da Cunha et al[22], 2025 economic/resource contrasts from Koh et al[23], 2024 (network meta-analysis). GRADE: Grading of recommendations assessment, development and evaluation; OR: Odds ratio; MD: Mean differences; RLR: Robotic liver resection; OLR: Open surgery; RoB: Risk of bias; NMA: Network meta-analysis.