As COVID ramps up again and skepticism about vaccines is spread, many people are wondering if it is worth the risk to have the shot, particularly cancer patients.
One of the key challenges with vaccines in general is to prove they don’t just work in healthy people, but also in patients with serious diseases, who often need the protection most. Excluding these patients from trials improves safety and the validity of trial results, but leaves doctors and patients scratching their heads about whether or not to vaccinate.
Now evidence is starting to be gleaned from clinical practice data to both guide the use of vaccines and improve their study.
A couple of studies that appear in the most recent JAMA Oncology added important new evidence. One looked at “Risk Factors for COVID-19–Related Hospitalization and Death in Patients With Cancer,” the other examined “COVID-19 Vaccine Booster Uptake and Effectiveness Among U.S. Adults With Cancer.”
Larry Han, PhD, of the department of public health and health sciences, Northeastern University, wrote an accompanying editorial on these reports, which were first published in July of this year. Inside Precision Medicine contacted Han about what the implications of these reports are.
Without evidence from randomized clinical trials (RCTs), carefully designed observational studies leveraging data from clinical practice are essential to estimating potentially heterogeneous vaccine effects in populations at higher risk for complications.
Patients with cancer may have immune responses affected by their treatment or disease progression, which could affect trial results. With booster doses, this becomes especially important, since risk-benefit tradeoffs may differ markedly across subgroups.
Han said patients with cancer should not automatically get a COVID vaccine. “Decisions should be individualized and made in consultation with patients’ physicians and family members, accounting for cancer type, stage, current treatment, and especially the degree of immunosuppression.”
He points out that, for example, patients on B-cell–depleting therapies may have attenuated immune responses and need additional considerations regarding timing, alternative prophylaxis, and boosters.
Han said some analysis challenges of including cancer patients in clinical trials are:
- Cancer patients are a heterogeneous group, so different treatments (chemo, immunotherapy, radiation) can all affect immune response.
- Their outcomes may be influenced by both cancer progression and COVID infection, making causal attribution harder.
- Standard models (like Cox proportional hazards) often break down because vaccine effects may wane over time or vary across subgroups.
- This points to the need for estimand-first causal survival analysis (e.g., restricted mean survival time, doubly robust estimation) to better define and estimate vaccine effects.
He suggests these challenges can be overcome through:
- Data fusion and transportability: Combining randomized trial results with real-world cancer cohorts using causal inference techniques to generalize findings.
- Federated causal inference: Allows multi-center studies without sharing raw patient data; adjusts for site-level heterogeneity to avoid misleading averages.
- “AI Hippocratic Oath” (no negative transfer): When combining datasets, the first rule is do no harm—don’t introduce bias. Ideally, we would like to leverage other data sources to increase the precision/reduce uncertainty in treatment effect estimates.
- Proximal causal inference: Use negative controls to detect and correct for hidden biases like “healthy vaccine effect.”
- Causal mediation analysis: Helps disentangle direct vaccine protection vs. indirect immune mechanisms, especially for patients on immunosuppressants.
- Accounting for interference: Vaccination may protect not just the patient but their close contacts (indirect effects).
The FDA and CDC, Han said, are already moving toward more targeted and individualized vaccination strategies, not universal boosters, precisely because risks and benefits differ. Most pivotal Phase III vaccine trials (e.g., Pfizer, Moderna) explicitly excluded cancer patients due to safety and internal validity concerns. As a result, the evidence base for vaccine efficacy in cancer populations has had to come from observational data and follow-up studies.
