Target trial emulation studies in oncology: a comprehensive analysis

Abstract

Background

Target trial emulation (TTE) has emerged as an important framework for generating comparative effectiveness real-world evidence. However, the methodological characteristics and clinical utility of this framework in oncology remain insufficiently characterized. This study systematically evaluated the characteristics of oncological emulation studies.

Materials and methods

We systematically searched PubMed, Embase, and Web of Science Core Collection databases from inception to December 22, 2023, to identify oncology-focused TTE studies. Study characteristics, TTE components, and consistency with corresponding randomized controlled trials (RCTs) were analyzed. Categorical variables were summarized as frequencies and percentages and compared using Chi-square tests in SPSS (version 27.0), with P-values <0 .05 considered statistically significant.

Results

A total of 54 TTEs were included. Most were conducted in the United States (22, 40.7%), and publications peaked in 2023 (17, 31.5%). Colorectal cancer (13, 24.1%) was the most frequently investigated cancer type. Studies predominantly used data from national databases (18, 33.3%). While all studies prospectively specified the target trial protocol and emulation procedures, only 1 (1.9%) study was prospectively registered, and most emulation studies (48, 88.9%) did not reference a reporting guideline. Among the 10 RCT-TTE pairs available for comparison, 6 (60.0%) pairs achieved agreement in both statistical significance and effect estimates.

Conclusion

This study provides a comprehensive analysis of TTE studies in oncology, highlighting their value for generating higher-quality real-world evidence. Future efforts should prioritize prospective protocol registration and standardize reporting to enhance the credibility and reproducibility of emulation studies.

Implications for Practice.

Target trial emulation (TTE) is a promising methodological framework for generating higher-quality real-world evidence in oncology, with substantial potential to inform clinical decision-making. Future TTE studies should prioritize prospective protocol registration and adhere to a well-designed reporting guideline to further enhance the transparency, credibility, and reproducibility of oncological research.

Introduction

Target trial emulation (TTE), proposed by Hernán and Robins in 2016, is a framework for designing and analyzing observational studies to evaluate the comparative effectiveness of interventions by emulating a hypothetical randomized controlled trial (RCT), the target trial, that would ideally answer the clinical question of interest.^1–3 This framework is implemented in two steps: first, specifying the protocol for the target trial, including eligibility criteria, treatment strategies, treatment assignment, follow-up, primary outcome, causal contrasts, and analysis plan, and second, emulating these components as closely as possible using observational data. By aligning observational study design with RCT principles, TTE can help reduce common biases in observational research and generate real-world evidence with greater clinical interpretability.^4,5

The need for reliable real-world evidence is particularly pressing in oncology, where many important comparative effectiveness questions cannot be fully addressed by RCTs due to ethical, logistical, financial, or feasibility constraints.⁶ In addition, oncology is characterized by complex, multi-line treatment pathways influenced by rapid disease progression and treatment-related toxicities, all of which may introduce time-varying confounding in conventional observational analyses.^7,8 In this context, the target trial framework provides a structured approach for addressing questions that remain unresolved by randomized trials or by conventional observational studies with conflicting findings.

For oncologists, emulation studies may help bridge the evidence gap between RCT findings and routine clinical practice by generating real-world evidence in patient populations that are often underrepresented in trials, such as older patients and those with comorbidities.⁶ They may also clarify treatment comparisons when direct randomized evidence is unavailable or difficult to obtain, thereby supporting clinical decision-making.⁹ In addition, TTEs may inform the design of future RCTs by refining eligibility criteria, identifying appropriate comparators, and prioritizing treatment strategies for prospective evaluation. Beyond clinical practice, well-designed emulation studies may also support evidence generation for pharmaceutical companies, particularly for innovative or high-cost therapies supported mainly by single-arm trials, by constructing external control arms from observational data, contextualizing treatment effects, and supporting market-access discussions.¹⁰ For regulatory stakeholders, TTEs may further support post-approval evidence generation by providing causal estimates at relatively low cost, thereby helping evaluate real-world effectiveness, safety, and generalizability in broader patient populations.^11–13

Despite the growing number of oncological emulation studies, their overall characteristics and application patterns have not been systematically analyzed. More importantly, it remains unclear whether, and to what extent, findings from oncological emulations align with results from corresponding RCTs. To fill these gaps, we comprehensively analyzed published oncological TTEs with the following objectives: (a) to systematically characterize their design and application pattern, (b) to evaluate core target trial components, and (c) to assess consistency between emulation outcomes and corresponding RCT results. By providing a comprehensive overview of current TTEs in oncology and addressing practical implementation challenges, we aim to offer clinical oncologists a valuable reference for interpreting and applying these findings to optimize clinical practice and decision-making.

Methods

Literature search

We searched PubMed, Embase, and Web of Science Core Collection databases from their inception until December 22, 2023. These databases were selected for their comprehensive coverage of published studies relevant to our research. The search strategy combined terms representing the concepts of “target trial emulation” OR “hypothetical trials” and “neoplasms” OR “tumor” OR “cancer.” The complete search strategy is available in Supplementary Material 1. Additionally, manual searches were conducted to ensure a comprehensive search.

Inclusion and exclusion criteria

The inclusion criteria were as follows: (a) retrospective observational studies that applied the target trial framework to emulate a hypothetical target trial, (b) studies that enrolled patients diagnosed with malignancies, evaluated interventions/exposures directly related to oncology, or assessed outcomes related to cancer development, and (c) studies written in English. The exclusion criteria included: (a) non-original research articles, including meeting abstracts, study protocols, methodological papers (methodological papers that aimed to illustrate the TTE framework, without addressing a specific, substantive clinical research question in oncology were excluded), editorials, commentaries, reviews, or theses, (b) articles unavailable in full-text.

Study screening and data collection

We used EndNote 9.0 software to manage the search records. Two independent reviewers (TCF and DZL) screened titles, abstracts, and full-texts of all studies. Studies meeting the eligibility criteria were included. Any disagreements were resolved by discussion.

Using a predesigned Excel form and guided by the target trial framework proposed by Hernán and Robins,¹ two reviewers independently extracted data. The process involved identifying and recording the components of the target trial protocol for each study. Any discrepancies were resolved by discussion. The following data were extracted and categorized into three domains: (a) Study characteristics, including publication year, country, data sources, sample size, conflicts of interest, funding, ethics approval, prospective study registration, reporting guidelines, and reasons why a TTE was conducted. (b) Target trial components, including eligibility criteria, treatment strategies, assignment procedures, follow-up periods, primary outcomes, causal contrasts, and analysis plans. (c) Methodological details, including the specification of the target trial protocol and emulation procedures (including their reporting format), sensitivity analyses, and reporting of potential unmeasured confounders.

Selection of corresponding RCTs and data extraction

For each included study, the two independent reviewers clarified its core research question and summarized the population, intervention, comparison, outcome, and study design (PICOS). We searched PubMed, Web of Science Core Collection, and ClinicalTrials.gov for published RCTs that addressed the same questions.

The RCT inclusion criteria were as follows: (a) study design was an RCT, (b) PIC elements were consistent with the corresponding emulation studies (minor variations were allowed if they were unlikely to alter the question), (c) the study reported outcomes as defined in the corresponding TTE, and (d) the study was cited or identified as the emulated target trial in the TTE study. RCTs were excluded if the results of outcomes could not be extracted or transformed for a comparable effect estimate in emulation studies. Any disagreements were resolved by discussion. The two reviewers then independently extracted data. For the primary outcome, both adjusted and unadjusted hazard ratios (HRs) with their 95% confidence intervals (CIs) were extracted from each TTE study and its corresponding RCT.

Statistical analysis

We assessed the following outcomes: (a) characteristics of published oncological TTEs, (b) core TTE components, and (c) consistency between TTE and corresponding RCT results.

Consistency was quantified using: (a) statistical significance agreement, defined as both the TTE and corresponding RCT being statistically significant in the same direction or both being statistically nonsignificant, (b) estimate agreement, defined as the TTE adjusted point estimate within the 95% CIs of the RCT result, and (c) standardized difference between the TTE and RCT effect estimates. The methodology adapted from those described by Wang et al.¹² Categorical variables were summarized as frequency counts and percentages and were compared using Chi-square tests. All analyses were performed using SPSS software (version 27.0), with P value <0 .05 considered statistically significant.

Results

Study characteristics

The literature search identified a total of 597 studies. After removing duplicates and screening titles, abstracts, and full-texts, 42 studies were identified. Additionally, 12 studies meeting the eligibility criteria were manually included, resulting in a total of 54 included studies.^{4,9,10,14–64} The study flowchart is presented in Figure 1.

Figure 1.

Study flow chart.

Among the included studies, most were conducted in the United States (22, 40.7%), and publications peaked in 2023 (17, 31.5%; Figure 2). Colorectal cancer (13, 24.1%) was the most frequently investigated cancer type. Regarding the study focus, 44 (81.5%) studies focused on treatment, 5 (9.3%) on screening, 2 (3.7%) on prevention, and 1 (1.9%) on medical facilities. In terms of study objective, 45 (83.3%) studies aimed to generate original evidence in the absence of RCTs, whereas only 9 (16.7%) sought to cross-validate or extend the findings of existing RCTs. The included studies predominantly used large-scale observational data sources, with national databases (18, 33.3%), Medicare databases (13, 24.1%), registry studies (10, 18.5%), and electronic health records (EHRs; 9, 16.7%). All studies prospectively specified the target trial protocol and clearly outlined the emulation procedures. However, only 1 (1.9%) study was prospectively registered, and most included studies (48, 88.9%) did not reference a reporting guideline. The detailed characteristics are shown in Table 1 and Supplementary Figure S1.

Figure 2.

Trend of publication years for the 54 included studies from 2015 to 2023.

Table 1.

Characteristics of the 54 included studies, 2015-2023.

Characteristics		Number of studies (%)
Country
	United States	22 (40.7)
	Multiple	7 (13.0)
	South Korea	4 (7.4)
	Sweden	4 (7.4)
	Denmark	3 (5.6)
	France	3 (5.6)
	Germany	3 (5.6)
	United Kingdom	3 (5.6)
	Canada	2 (3.7)
	China	1 (1.9)
	Netherlands	1 (1.9)
	Thailand	1 (1.9)
Cancer type
	Colorectal cancer	13 (24.1)
	Prostate cancer	9 (16.7)
	Hepatocellular carcinoma	8 (14.8)
	Pancreatic cancer	5 (9.3)
	Bladder cancer	4 (7.4)
	Multiple	4 (7.4)
	Breast cancer	3 (5.6)
	Esophageal cancer	1 (1.9)
	Intrahepatic cholangiocarcinoma	1 (1.9)
	Lung cancer	1 (1.9)
	Oesophagogastric adenocarcinoma	1 (1.9)
	Ovarian cancer	1 (1.9)
	Renal cell carcinoma	1 (1.9)
	Relapsed/refractory multiple myeloma	1 (1.9)
	Spinal metastasis	1 (1.9)
Data sources
	National databases	18 (33.3)
	Medicare databases	13 (24.1)
	Registry study	10 (18.5)
	EHRs	9 (16.7)
	Clinical trials	3 (5.6)
	Multiple data sourcesa	1 (1.9)
Conflicts of interests
	Noneb	32 (59.3)
	Multiple relationshipsc	12 (22.2)
	Employment relationship	3 (5.6)
	Research support	3 (5.6)
	Consulting relationship	2 (3.7)
	Not mentioned	1 (1.9)
	Otherd	1 (1.9)
Funding sources
	Other institutional and foundation fundinge	18 (33.3)
	Not mentioned	12 (22.2)
	Government funding	10 (18.5)
	Multiple funding sourcesf	9 (16.7)
	Corporate sponsorship	3 (5.6)
	None	2 (3.7)
Ethics approval
	Ethics approved	34 (63.0)
	Exempt from ethics approval	16 (29.6)
	Not mentioned	4 (7.4)
Study registration status
	Yes	1 (1.9)
	No	53 (98.1)
Reporting guideline
	Not mentioned	48 (88.9)
	STROBE	2 (3.7)
	STROBE and ISPOR	2 (3.7)
	ISPOR	1 (1.9)
	RECORD	1 (1.9)
Specification of the emulation procedures
	Yes	54 (100.0)
	No	0 (0)
Specification of the target trial protocol
	Yes	54 (100.0)
	No	0 (0)
Reporting format of the protocol
	Table	34 (63.0)
	Text	20 (37.0)
Sensitivity analyses
	Yes	45 (83.3)
	No	9 (16.7)
Reporting of potential unmeasured confounders
	Yes, without E-value	50 (92.6)
	Yes, with E-value	4 (7.4)

Abbreviations: EHR, electronic health record; ISPOR, International Society for Pharmacoeconomics and Outcomes Research; RCT, randomized controlled trial; RECORD, reporting of studies conducted using routinely collected data; STROBE, strengthening the reporting of observational studies in epidemiology; TTE, target trial emulation.

^aStudies using multiple data sources, including EHRs, national databases, health insurance databases, or clinical research data.

^bStudies reporting no conflicts of interest.

^cStudies reporting multiple types of conflicts of interest, including employment, research support, consulting relationships, or editorial affiliations.

^dStudies reporting other conflicts of interest not categorized as employment, research support, consulting relationships, or editorial affiliations.

^eFunding from institutional, foundation, or other sources excluding government and corporate sources.

^fFunding from multiple sources, including government, corporate, institutional, or foundation sources.

Components of TTE

Following the target trial framework proposed by Hernán and Robins,¹ we systematically reviewed the key protocol components across the 54 included studies. Regarding treatment strategies, comparisons were categorized into two types: presence-absence intervention comparisons (34, 63.0%) and intervention modality/detail variations (20, 37.0%). The investigated strategies contained: pharmacology (26, 48.1%), surgery (9, 16.7%), cross-modal therapies (5, 9.3%), screening strategies (5, 9.3%), radiotherapy (4, 7.4%), lifestyle modifications (2, 3.7%), supportive care (2, 3.7%), and medical facility utilization (1, 1.9%). The most frequently assessed primary outcomes were mortality (35, 64.8%) and cancer incidence (13, 24.1%), with fewer studies assessing patient well-being (5, 9.3%) or pathological response (1, 1.9%). Follow-up periods ranged from 30 days to 15 years, depending on the specific research question. Detailed information regarding bias adjustments is presented in Supplementary Table S1.

Consistency assessment

Among the 54 included studies, nine TTEs corresponded to published RCTs with comparable outcome data. One TTE study emulated two separate RCTs, resulting in a total of 10 RCT-TTE comparison pairs. Among these pairs, 6 (60.0%) achieved agreement in both statistical significance and effect estimates (dual consistency). All 10 (100.0%) pairs achieved estimate agreement, whereas 6 (60.0%) achieved statistical agreement. Detailed information is presented in Table 2.

Table 2.

The consistency between TTE and corresponding RCT results.

Study	Effect estimates (95% CIs)					Agreement metrics
	Trial name/NCT number	RCT	TTE adjusted	TTE crude	Standardized difference	Statistical significance	Estimate	Standardized- difference
164	QUASAR	1.02 (0.70-1.48)	0.95 (0.85-1.04)	1.00 (0.89-1.12)	0.36	SA	EA	SD
264	Erlotinib trial	0.96 (0.74-1.24)	1.04 (0.86-1.42)	1.08 (0.93-1.25)	−0.42	SA	EA	SD
343	NCT01734798	0.61 (0.33-1.11)	1.11 (0.82-1.51)	1.44 (1.07-1.92)	−1.90	SA	EA	SD
428	EORTC 30881	1.02 (0.80-1.29)	1.13 (1.07-1.20)	1.26 (1.20-1.33)	−1.38	N	EA	SD
541	IDEA	0.96 (0.85-1.08)	0.96 (0.43-2.14)	NA	0.01	SA	EA	SD
614	EORTC 55955	1.01 (0.82-1.25)	1.12 (0.96-1.28)	NA	−0.78	SA	EA	SD
715	NCT00002575	0.91 (0.68-1.21)	0.84 (0.79-0.89)	NA	0.68	N	EA	SD
854	NCT02491983	1.00 (0.68-1.48)	1.07 (0.86-1.35)	1.24 (1.02-1.51)	−0.30	SA	EA	SD
957	NCT00309985	0.72 (0.59-0.89)	0.71 (0.42-1.19)	0.68 (0.45-1.03)	0.09	N	EA	SD
1059	NCT00028990	0.88 (0.74-1.05)	0.84 (0.77-0.92)	0.89 (0.82-0.98)	0.71	N	EA	SD

Abbreviations: CIs, confidence intervals; EA, estimate agreement, with TTE adjusted point estimates falling within the 95% CIs of the corresponding RCT result; SA, statistical significance agreement, with both the TTE and corresponding RCT being statistically significant in the same direction or both being statistically nonsignificant; N, No; NA, not applicable; NCT, national clinical trial; RCT, randomized controlled trial; SD, standardized-differences agreement, defined by standardized differences |Z| < 1.96; TTE, target trial emulation.

Discussion

Overview of TTE in oncology

Our study demonstrated that the target trial framework has become an important approach for filling evidence gaps in oncology. Most included studies focused on therapeutic interventions with long-term survival outcomes and used large-scale real-world data. These studies were particularly relevant to clinical, regulatory, and health-technology decision-making when randomized evidence is unavailable or insufficient.

This framework is especially valuable for addressing oncology questions that are difficult to answer in randomized trials, including studies of underrepresented populations, head-to-head comparisons of treatment regimens, complex treatment strategies, long-term outcome assessments, and drug repurposing.^{4,40,42,44,65,66} For example, Medicare data have been used to emulate target trials comparing adjuvant fluorouracil versus observation for stage II colorectal cancer and evaluating the addition of erlotinib to gemcitabine for advanced pancreatic cancer in older populations.⁶⁴ TTEs also enable head-to-head comparisons of routine clinical regimens. An example is the comparison between FOLFIRINOX (folinic acid, fluorouracil, irinotecan, and oxaliplatin) and GN (gemcitabine plus nab-paclitaxel) for advanced pancreatic carcinoma, filling an important comparative-effectiveness gap left by separate trials.⁹ Beyond these examples, this method may help evaluate treatment sequences, cycle numbers, dosing, dynamic treatment strategies influenced by disease progression and patient adherence, long-term outcomes, and potential drug repurposing opportunities.^40,44,65,66

Overall, emulation studies should be regarded as structured complements to RCTs rather than substitutes for them. Their value lies in translating clinically important but otherwise difficult-to-test questions into explicitly specified target trial protocols that can be implemented using observational data.

Challenges of TTE in oncology

Despite its utility, current applications of TTE in oncology face several challenges. First, data relevance and reliability determine whether the emulation can appropriately answer the intended clinical question.^67,68 Oncological research requires not only demographic information but also a broad range of clinical variables, such as medical history, tumor characteristics, and comorbidities. However, national databases, the primary data sources for these studies, often suffer from missing clinical variables and incomplete treatment records, introducing unmeasured biases.⁶⁹ Future emulation studies should prioritize data standardization and integrate multi-source linked data to enhance data quality.

Second, clearly defining “time zero” is essential in target trial emulation.¹ Failure to synchronize eligibility, treatment assignment, and follow-up can introduce immortal time bias.⁸ Researchers should establish standardized, clinically meaningful “time zero” events (e.g., date of diagnosis, first treatment, or enrollment) and conduct sensitivity analyses to assess the robustness of the results.

Finally, the transparency and reproducibility of emulation studies remain insufficient. Our study showed that prospective registration and use of reporting guidelines were uncommon in oncological TTEs. Pre-specifying the target trial protocol before outcome analysis can enable independent appraisal of analytic decisions, reduce publication bias, and improve study credibility. Moreover, standardized reporting guidelines are still lacking despite the growing application of TTE.⁷⁰ Transparent reporting of the study protocol and its emulation process is crucial to enhance the interpretability and utility of findings.^70–73 Recently, the Transparent Reporting of Studies Emulating a Target Trial (TARGET) reporting guideline represents a promising tool to standardize TTE reporting.⁷⁴

Guidance for clinical practice

Oncologists should interpret TTEs as well-designed observational studies, not as true RCTs, since these studies can never completely replicate blinding and randomization. Specifically, when assessing such studies, clinicians should rigorously evaluate the data source quality, the definition of “time zero,” analysis plans, and reporting transparency. Findings from emulations with small sample sizes, low emulation fidelity, or extrapolated beyond the target population should not be applied to clinical practice.¹ Only through such rigorous appraisal can emulation studies effectively fulfill their role as a key complement to RCTs, successfully integrating real-world evidence into cancer prevention, screening, and treatment.

Limitations

This study provided a comprehensive overview of current oncological emulation studies; however, several limitations should be acknowledged. First, our assessment of TTE characteristics and methodological quality was based on reported information. Incomplete or suboptimal reporting may have led to misclassification of certain design features or analytic approaches. Second, the number of TTEs with corresponding published RCTs was limited, which constrained the statistical power of the consistency assessments.

Conclusion

In conclusion, the present study provides a comprehensive analysis of the application of the target trial framework within the oncology literature. Our findings demonstrate that the TTE framework should be considered a promising methodology for generating high-quality real-world evidence, particularly when RCTs are not feasible. Future efforts should prioritize the prospective registration of emulation studies to further bolster the credibility and reproducibility of real-world research.

Author contributions

Chenfeng Tan (Data curation, Formal analysis, Writing—original draft), Zelei Dai (Data curation, Formal analysis, Writing—original draft), Renjie Zhao (Data curation), Kefan Li (Data curation), Bingran Zhang (Data curation), Ying Wang (Data curation), Nian Li (Conceptualization, Methodology, Supervision), Lei Liu (Conceptualization, Methodology, Supervision), and Yonggang Zhang (Conceptualization, Supervision, Writing—review & editing)

Supplementary material

Supplementary material is available at The Oncologist online.

Funding

This study was supported by the Sichuan Science and Technology Program (2023NSFSC1556 and 2023YFQ0071) and the Medical Quality Evidence-based Management Research Project of the Hospital Management Institute of the National Health Commission in China (YLZLXZ23K009 and YLZLXZ23G042).

Conflicts of interest

None declared.

Data availability

The data is available upon reasonable request to the corresponding author.