Restricting Network Meta-Analysis to Primary Colon Cancer Resection
Yes, you can and often should restrict your network meta-analysis to patients undergoing primary colon cancer resection only, as this approach strengthens the validity of your findings by ensuring transitivity—the assumption that patient populations are comparable across studies.
Methodological Rationale
The PRISMA-NMA guidelines 1 explicitly address this "lumping versus splitting" decision. Restricting to a homogeneous population (primary resectable colon cancer only) maximizes the plausibility of transitivity, which is the fundamental assumption that patient characteristics and effect modifiers are comparable across your treatment network.
Key Considerations for Restriction:
Population Homogeneity:
- Patients undergoing primary colon cancer resection represent a distinct clinical entity with different prognosis, treatment goals, and outcomes compared to:
Effect Modifier Control:
- By restricting to primary resectable colon cancer, you eliminate major effect modifiers including:
- Tumor location (colon vs. rectum have fundamentally different treatment paradigms)
- Disease stage (resectable vs. metastatic)
- Treatment intent (curative vs. palliative)
- Surgical approach variations
Evidence Supporting Restriction
The literature demonstrates that mixing populations introduces heterogeneity that undermines network validity:
- Rectal cancer requires distinct management: Neoadjuvant chemoradiotherapy is standard for rectal cancer 2 but not for colon cancer, making these populations non-exchangeable
- Metastatic disease changes the treatment question entirely: Studies of unresectable mCRC 3, 5 address fundamentally different interventions (whether to resect primary tumor, later-line systemic therapies) versus adjuvant treatment after curative resection 6
- Surgical context matters: Studies mixing elective and emergency surgery 4 introduce confounding that obscures treatment effects
Practical Implementation
When defining your eligibility criteria, explicitly state:
- Include: Patients with histologically confirmed primary colon adenocarcinoma undergoing curative-intent surgical resection
- Exclude: Rectal cancer (typically defined as <15cm from anal verge), metastatic disease at presentation, emergency surgery, recurrent disease
Document your rationale: The PRISMA-NMA guidelines 1 emphasize that "arguments in favor of defining the evidence base in a way that maximizes the plausibility of transitivity" should be clearly articulated. State that restriction to primary resectable colon cancer ensures comparable baseline risk, treatment intent, and outcome measurement across your network.
Important Caveats
Trade-off with network connectivity: Restriction may reduce the number of available studies and potentially disconnect parts of your treatment network. However, validity trumps network size—a smaller network with defensible transitivity is superior to a large network with questionable comparability 1.
Stage-specific considerations: Even within resectable colon cancer, you may need to consider whether to further stratify by stage (high-risk stage II vs. stage III), as treatment effects may differ 6. The evidence suggests 3-month versus 6-month oxaliplatin-based regimens show different relative efficacy in low-risk versus high-risk stage III patients 6.
Transparency in reporting: If you exclude studies of rectal cancer or metastatic disease, clearly report the number of excluded studies and reasons in your PRISMA flow diagram to allow readers to judge the comprehensiveness of your evidence base 1.