Table 8. Quality assessment.

Category	Outcome
	Resection Rate	Exploration Rate	Response Evaluation	Toxicity	Morbidity	Mortality	Median Survival	1 y/2 y Survival
Number of studies (range)	35–111	32–88	2–82	22–63	22–50	30–85	19–70	11–54
Study design	Phase I–II trials	Phase I–II trials	Phase I–II trials	Phase I–II trials	Phase I–II trials	Phase I–II trials	Phase I–II trials	Phase I–II trials
	Cohort studies	Cohort studies	Cohort studies	Cohort studies	Cohort studies	Cohort studies	Cohort studies	Cohort studies
	Case series	Case series	Case series	Case series	Case series	Case series	Case series	Case series
Limitations	No serious limitation	No serious limitation	No serious limitationa	No serious limitation	No serious limitationb	No serious limitationb	No serious limitationc	No serious limitationd
Inconsistency	Serious inconsistencye	Serious inconsistencye	Serious inconsistencye	No serious inconsistencyf	No serious inconsistencyg	No serious inconsistencyh	Serious inconsistencye	Serious inconsistencye
Indirectness i
Imprecisionj	No serious imprecisionk	No serious imprecisionk	No serious imprecisionk	No serious imprecisionl	Serious imprecisionm	Serious imprecisionn	Serious imprecisiono	Serious imprecisiono
Other considerations							Studies weighted by number of initial patients	Studies weighted by number of initial patients
Quality	++	++	++	+++	++	++	+	+
Importance	Critically important	Important	Important	Important	Important	Critically important	Critically important	Critically important

The quality assessment for the indicated outcome parameters was carried out according to the grade profiler as described in the Methods section. +, very low; ++, low; +++, moderate.

^aSeparated evaluation of response categories, several evaluation criteria, proportions experiencing each type of response are not independent.

^bBased only on resected patients.

^cAssumption of uniform (exponential) distributed survival times.

^dDifferent trials for estimating 1 y/2 y survival.

^eHigh heterogeneity which could not be sufficiently explained by potential sources of variation within the meta-regression analysis.

^fObvious heterogeneity partly explained by different chemotherapy treatment regimes within the trials.

^gObvious heterogeneity partly explained by differences in mean patient age and study design between the trials.

^hNo considerable heterogeneity obvious from the data.

ⁱSince no direct comparison was feasible for any considered outcome measurement, indirectness has to be fixed as “very serious” for the entire topic of the investigation.

^jFor the same reason no particular measure of precision (e.g. confidence intervals) was available for any single trial considered; therefore, imprecision is rather crude assessed in reference to the number of study participants within any meta-analysis.

^kSubstantial sample sizes per study: median (IQR): 30 (18 to 47).

^lSubstantial sample sizes per study: median (IQR): 29 (19 to 39).

^mInsufficient sample sizes per study: median (IQR): 16 (7 to 25).

ⁿInsufficient sample sizes per study: median (IQR): 13 (6 to 25).

^oNo sufficient information about patients at risk.