How direct-to-tumor cancer treatments work
Intratumoral means a clinician places a needle into a tumor and delivers a treatment there. Intralesional is often used similarly, especially for visible skin or node lesions. The route says where the dose starts; it does not guarantee that the drug stays there or that distant tumors respond. This approach needs a measurable, reachable lesion. It cannot be injected into microscopic disease that has no identifiable target.
Start with three separate questions
- What goes into the tumor? A virus, antibody, immune signal, gene payload, chemotherapy, bacterium, or engineered cell can all be injected. The word “intratumoral” does not identify the treatment.
- What happens to the injected lesion? Direct cell injury or local immune activation may shrink it. An injected lesion's response is the most direct outcome to measure.
- What happens elsewhere? Immune cells may travel to uninjected tumors, but a whole-body response must be measured separately. A local response or a blood immune marker alone does not prove distant tumor control.
An injection can be paired with systemic treatment. For example, a trial may inject an experimental drug into a tumor and give pembrolizumab (Keytruda) intravenously. Its title might say “intratumoral treatment plus pembrolizumab,” even though pembrolizumab itself was not injected. The route for each drug must be checked in the trial methods. KEYNOTE-890 registry; MK-1454 registry.
The major approaches
| What is injected | Simple mental model | Human example | What the example establishes |
|---|---|---|---|
| Oncolytic virus | A virus infects tumor cells and can also alert the immune system. | T-VEC, an engineered herpes virus, is FDA-approved for injection into certain recurrent melanoma lesions. In a single-arm TNBC study, T-VEC was added to chemotherapy before surgery. | Local virus treatment is feasible; the TNBC study cannot isolate T-VEC's benefit from chemotherapy. The measles/VSV Halassy case used different viruses and also involved surgery and postoperative trastuzumab. |
| Checkpoint antibody | Blocks an immune-cell “brake” inside the treated lesion. | Ten patients with high-risk DCIS received intratumoral pembrolizumab itself plus mRNA-2752 in a phase 1 cohort. | A real direct-Keytruda precedent: 8/10 had a reported response, including 3 complete responses. It was small, nonrandomized, involved noninvasive DCIS, and tested a combination, so it does not establish efficacy in invasive TNBC or pembrolizumab alone. |
| Cytokine gene payload | Gives tumor cells instructions to produce an immune signal nearby. | TAVO delivers plasmid IL-12 into TNBC lesions, followed by electroporation to help cells take it up; KEYNOTE-890 combines it with IV pembrolizumab. | Immune changes and early clinical signals, not a proven survival gain. The TNBC combination record's current status is unknown after its last 2023 update. |
| Innate immune stimulant | Sounds an immune alarm, such as STING or TLR activation. | Intratumoral MIW815/STING in 47 patients produced one confirmed partial response despite evidence of systemic immune activation. BO-112 plus IV pembrolizumab had a 25% response rate among 40 evaluable anti-PD-1-resistant melanoma patients in a single-arm phase 2 study. | Immune activation and tumor benefit are different outcomes. These mixed or non-breast cohorts do not establish TNBC benefit. |
| Tumor-directed chemotherapy formulation | Delivers cytotoxic drug within the lesion to concentrate local injury. | INVINCIBLE-4-SAKK randomizes early TNBC patients to an injection of INT230-6 into the breast tumor before standard neoadjuvant immunochemotherapy, or standard treatment alone. | A useful comparative trial design, recruiting as of September 25, 2026; efficacy results are not yet established by the registry. |
| Engineered bacteria | Modified bacteria can grow in low-oxygen tumor regions and cause local injury and inflammation. | In a 16-patient phase 1b study, Clostridium novyi-NT spores were injected into tumors while pembrolizumab was given IV. | Early safety and response information in mixed advanced cancers. One grade 3 abscess was dose limiting; this is not routine cancer treatment. |
Other experimental payloads include dendritic cells, CD40 agonist antibodies, TLR agonists, mRNA particles, and locally injected microdoses used to test drug effects. A trial listing demonstrates that an approach is being studied; it does not demonstrate efficacy. See the dated evidence and trial map for examples and exact routes.
Why local delivery is appealing—and what can go wrong
The hope is high exposure in a tumor, less exposure elsewhere, and an immune response that reaches untreated sites. Each of those is a hypothesis to measure, not an automatic property of injection. Drug can enter the bloodstream; fever, inflammation, immune toxicity, infection, and injection-site injury remain possible. Needle access can be unsafe near vessels or organs. Some trials require both an injectable lesion and a separate uninjected measurable lesion so distant effects can be tested. MIW815 phase 1; Clostridium phase 1b.
Pathologic complete response (pCR) means no residual invasive cancer in a defined surgical specimen after preoperative treatment. It is different from shrinkage on imaging and from long-term survival. In a combination trial, even a high pCR rate cannot tell us how much the injected agent added without a suitable comparator. An abscopal or systemic effect means a distant, uninjected lesion responds; it needs separate lesion measurements, not just immune markers. The Halassy and T-VEC evidence review shows why these distinctions matter.
A five-question checklist for reading a study
- Which drug was injected, and which was given IV, orally, or by another route?
- Was the disease DCIS, operable invasive cancer, locally recurrent cancer, or metastatic cancer?
- What outcome was measured: injected-lesion shrinkage, uninjected-lesion response, pCR, recurrence, or survival?
- Did a control arm receive the same background therapy without the injection?
- Was treatment part of a registered trial with controlled manufacturing, dosing, monitoring, and a plan for surgery or subsequent care?
The Elicit search archive records the discovery searches. The linked papers and registries above govern the claims. This page explains a research area; it does not imply a current treatment option for a person without an injectable lesion or recommend changing a care plan.