NK Cell Therapy: What the Evidence Shows

Medically reviewed by a licensed physician · Last reviewed 2026-08-05

Before you read further. Nothing here is a reason to delay, interrupt, or decline treatment recommended by your oncology team. If you are considering any additional therapy, discuss it with the physician managing your cancer first.

The short version

If you read nothing else, read this.

NK cells are part of your immune system. They find and kill damaged cells, including some cancer cells. This is normal biology, not a laboratory invention.

NK cell therapy means taking these cells, growing more of them, and putting them back. Doctors have been studying this for decades.

“NK cell therapy” is not one treatment. There are many different versions. They use cells from different sources, prepared in different ways. A result from one version does not tell you much about another. This is the single most misunderstood thing in this field.

Some versions have worked in some blood cancers. In one study, a genetically engineered version was given to 37 patients with lymphoma and leukemia, and about half responded. That result belongs to that specific product — it does not transfer to other versions or to other cancers.

For solid tumors — lung, breast, liver, pancreas and others — the research is at an earlier stage. Studies have shown tumors shrinking in some patients, and the treatment has generally been well tolerated. What has not yet been done is the large comparison study that would show whether patients live longer. Early evidence is not the same as evidence that something failed — it means the question is still open.

Side effects have generally been mild. Fatigue is the most common. But few side effects does not mean it works — a treatment that does nothing also causes no harm.

No NK cell therapy is approved by the FDA to treat cancer. It remains under study.

The number of cells does not tell you if a treatment is strong. Neither does the word “activated.” Ask what else is in the product.

What comes next

The rest of this page explains all of the above in detail — the biology, the actual study results with their numbers, why solid tumors are harder, what to ask any provider, and which claims the evidence does not support.

It is long, because the honest answer is not short. You do not have to read it all at once, and you can bring it to your own doctor.

What natural killer cells are

NK cells are part of your innate immune system — the part that responds immediately, without needing to be trained first. Roughly 5 to 15% of the lymphocytes circulating in your blood are NK cells.

Unlike T cells, which must first be shown a specific target, NK cells act on a balance of signals. Activating receptors tell them to kill; inhibitory receptors tell them to stand down. When activating signals outweigh inhibitory ones, the NK cell forms a connection with the target and destroys it.

How they kill cancer cells

Understanding these mechanisms helps you evaluate claims, because most marketing describes mechanisms and lets you infer results.

Perforin and granzymes. The main pathway. The NK cell attaches to the target, perforin (a protein that opens a passage through the target cell’s outer layer) allows granzymes (enzymes that trigger self-destruction) to enter and shut the cell down. This requires physical contact.

Death receptors. NK cells carry molecules (FasL and TRAIL) that activate self-destruct programs in cells displaying the matching receptors. Tumors vary in how sensitive they are to this, and can become resistant.

“Missing self.” Healthy cells display HLA class I molecules — surface markers that identify a cell as your own — and these switch NK cells off. Many tumors reduce HLA to hide from T cells — and that same reduction can make them visible to NK cells. This is the core rationale for the whole field.

Stress ligand recognition. Tumors under genomic stress display molecules that NK receptors like NKG2D detect. But tumors fight back: they can shed these ligands, release them in soluble form, or package them in vesicles that blunt NKG2D function.

Antibody-dependent killing (ADCC). When a therapeutic antibody coats a tumor cell, the NK cell’s CD16 receptor recognizes it and kills the marked cell. This is why NK cells are studied alongside antibodies like rituximab, trastuzumab and daratumumab. Note what does the targeting here: the antibody.

Cytokine signaling. Cytokines are chemical messengers immune cells use to communicate. NK cells release interferon-gamma and other cytokines that recruit dendritic cells, macrophages and T cells — so their potential effect extends beyond direct killing.

All of this is well established in laboratory and animal models. Biological plausibility is where research starts. The rest of this page is about what happened in patients.

⚠️ The most important thing on this page

“NK cell therapy” describes radically different products. They are not interchangeable, and a result from one tells you very little about another.

They also differ in expansion method, cytokines used, dose, whether lymphodepletion was given first, and what other treatments were combined.

This means: results from a cord-blood-derived CAR-NK engineered to express IL-15, given after chemotherapy conditioning to a patient with B-cell lymphoma, cannot be used to support an infusion of unmodified NK cells in a patient with pancreatic or breast cancer. Different product, different mechanism, different disease.

When you see NK cell results cited anywhere — including here — ask which product produced them.

Where the evidence is strongest: CAR-NK in B-cell cancers

This is the most convincing clinical evidence in the field, and it is worth understanding in detail.

Researchers used NK cells from umbilical cord blood, engineered to carry a receptor targeting CD19, to produce IL-15, and to include a safety switch. Patients first received lymphodepletion — chemotherapy given to reduce their existing immune cells, which makes room for the new ones to expand.

The 2024 update in Nature Medicine included 37 patients with relapsed or refractory CD19-positive cancers:

• Overall response rate: 48.6% at day 30 and day 100
• One-year overall survival: 68%
• Progression-free survival: 32%
• No cytokine release syndrome (a dangerous immune overreaction), no neurotoxicity (confusion or seizures), and no graft-versus-host disease (donor cells attacking the patient’s tissues)

Response varied substantially by diagnosis: 100% in low-grade non-Hodgkin lymphoma, 67% in chronic lymphocytic leukemia without transformation, and 41% in diffuse large B-cell lymphoma. Those groups were small, so the percentages should not be read as reliable rates.

How to read this correctly. A 48.6% response rate is real clinical activity in patients who had run out of options. But 32% progression-free survival at one year shows that control was neither universal nor necessarily lasting. And the study was phase I/II, single-center, with no randomized comparison group.

And this is essential: these results belong to that specific engineered product and protocol. They are not evidence for NK cell therapy generally.

Source — phase 1/2 trial, 2024. Marin D, et al. Safety, efficacy and determinants of response of allogeneic CD19-specific CAR-NK cells in CD19+ B cell tumors. Nature Medicine. PubMed 38238616 · doi:10.1038/s41591-023-02785-8

Acute myeloid leukemia

AML has the longest history of NK cell research, and there are real signals.

Cytokine-induced memory-like NK cells — briefly exposed to specific immune signals (IL-12, IL-15 and IL-18) before being given — showed enhanced anti-leukemia activity and produced responses in early studies. In pediatric and young-adult patients who relapsed after transplant, one phase 1 study reported complete remission in four of eight evaluable patients at day 28, two of them lasting beyond three months. Those patients also received donor lymphocyte infusions alongside the NK cells, so what each contributed cannot be separated.

These are small studies in highly selected patients, often alongside transplant or lymphodepletion, without control groups. They justify continued research; they do not establish a standard of care.

A consistent finding across AML studies: NK cells appear to work better against low disease burden — residual disease, or after the immune environment has been altered by conditioning — than against bulky tumors.

Sources — two phase 1 trials. Romee R, et al. Cytokine-induced memory-like natural killer cells exhibit enhanced responses against myeloid leukemia. Science Translational Medicine, 2016. PubMed 27655849 · doi:10.1126/scitranslmed.aaf2341

Bednarski JJ, et al. Donor memory-like NK cells persist and induce remissions in pediatric patients with relapsed AML after transplant. Blood, 2022. PubMed 34871371 · doi:10.1182/blood.2021013972

A trial where it did not work

A phase II trial at St. Jude Children’s Research Hospital studied 21 children with acute myeloid leukemia in first remission who received donor NK cells as consolidation therapy, compared against 55 children who received chemotherapy alone.

The cells behaved as expected: they expanded in 17 of the 21 patients (81%), with no major adverse effects. But the treatment did not reduce relapse or improve survival compared with chemotherapy alone.

The authors noted explicitly that this result does not rule out usefulness during other phases of treatment, or in combination with other immunotherapies.

This trial is on this page for a reason. A treatment can work biologically, be safe, and still not change outcomes in a particular setting — and being able to recognize that is more useful to you than any single percentage. Negative trials are a normal part of developing any therapy; chemotherapy, radiation and immunotherapy all have many. What this one tells you is something specific about dose and persistence, which later research has tried to address.

Source — phase II trial, 2019. Nguyen R, et al. A phase II clinical trial of adoptive transfer of haploidentical natural killer cells for consolidation therapy of pediatric acute myeloid leukemia. Journal for ImmunoTherapy of Cancer. PubMed 30894213 · doi:10.1186/s40425-019-0564-6

Solid tumors

This is where most patients seeking treatment are, and where the research is at an earlier stage than in blood cancers.

NK therapies have been studied in lung, liver, ovarian, breast, pancreatic, kidney, prostate, colon cancer, melanoma, sarcoma and glioblastoma. This is an active field: hundreds of trials are underway.

What has been observed. A 2024 systematic review of unmodified NK therapies in solid tumors, covering 31 trials and roughly 600 patients, found measurable signals of activity — tumors shrinking in some patients — alongside a safety profile the authors described as acceptable. Some patients experienced temporary stabilization of disease. These are real observations in real patients.

What has not been established. The same review found substantial variation between studies and stated explicitly that controlled trials are needed. Nearly all of these studies are small, single-arm, and early-phase, using different products — which means they can show that something happened, but not that the NK cells caused it or that patients lived longer as a result.

The highest reported figure in that review was a 72.3% objective response rate in hepatocellular carcinoma, which requires three pieces of context:

• It was not NK cells alone — those patients received other local treatments simultaneously
• “Objective response” means the tumor shrank, not that patients lived longer
• The trials had no control group, so there is no way to know how those patients would otherwise have fared

Stated plainly: for most solid tumors, safety and feasibility are better established than effectiveness. There is not yet adequate evidence that unmodified NK cells improve overall survival compared with standard care — and that is different from evidence that they do not.

One pattern does appear across the research, and it is worth understanding because it may be relevant to your situation. NK cells appear to work better against lower disease burden — smaller tumors, residual disease after other treatment, or settings where the immune environment has been altered first — than against large, established tumors. This is consistent with what is known about the obstacles described below: the more tumor there is, the harder it is for NK cells to reach it, survive in it, and outnumber it.

Whether that pattern applies to your case is a clinical question, and it is one worth asking a physician directly.

Source — systematic review and meta-analysis, 2024. Park H, et al. Efficacy and safety of natural killer cell therapy in patients with solid tumors. Frontiers in Immunology. PubMed 39478866 · PMC11522797 · doi:10.3389/fimmu.2024.1454427

Why solid tumors are so much harder

Seven documented obstacles, and they explain the gap between blood cancers and solid tumors:

Trafficking. After IV infusion, many NK cells lodge initially in the lungs, liver and spleen. They do not necessarily reach the tumor in useful numbers.

Penetration. Solid tumors contain dense stroma, fibrosis, abnormal blood vessels, high internal pressure, and low-oxygen regions that block immune cell access.

Immunosuppression. Tumors surround themselves with substances that switch immune cells off — TGF-beta, adenosine, low oxygen (hypoxia), lactate and others. These reduce NK cytotoxicity and survival inside tumors.

Antigen heterogeneity. A targeted product only recognizes cells carrying its target. Tumors containing negative subclones can escape.

Limited persistence. NK cells often disappear before eliminating the tumor, producing transient reduction without durable control.

Checkpoint inhibition. Immune cells carry built-in brakes — receptors called NKG2A, TIGIT and PD-1 — that tumors can press to shut them down. In lung cancer, NK cells found inside tumors carried less perforin and granzyme than those in circulating blood.

Tumor-induced dysfunction. Tumors release soluble factors and vesicles that reduce NKG2D and disrupt the connection NK cells need to make.

Source — review, 2025. Balkhi S, et al. CAR-NK cell therapy: promise and challenges in solid tumors. Frontiers in Immunology. PubMed 40260240 · doi:10.3389/fimmu.2025.1574742

About autologous cells specifically

Because “we use your own cells” is a common selling point, this deserves its own section.

Autologous cells avoid donor-recipient incompatibility. But they carry a specific limitation: if the tumor retains enough of your own HLA, your NK cells’ inhibitory receptors may prevent them from attacking it — even after expansion and activation outside the body. Self-tolerance does not disappear because cells were cultured.

Additionally, NK cells taken from cancer patients may already be exhausted or impaired by the disease, prior chemotherapy, systemic inflammation, or age.

Historically, autologous NK cells have been shown to be safely administered, but their anti-tumor effect has been modest or inconsistent, particularly in solid tumors. That is a substantial part of why the field moved toward healthy donors, memory-like cells and genetically modified products.

“We use your own immune system to fight cancer” describes a procedure. It does not demonstrate effectiveness, and personalization does not by itself make a product more potent.

Sources — two reviews. Oh S, et al. Natural killer cell therapy: a new treatment paradigm for solid tumors. Cancers, 2019. PubMed 31614472 · doi:10.3390/cancers11101534

Lizana-Vasquez GD, et al. The application of autologous cancer immunotherapies in the age of memory-NK cells. Frontiers in Immunology, 2023. PubMed 37205105 · doi:10.3389/fimmu.2023.1167666

Why cell count does not tell you potency

The general version of this — why an advertised quantity is not a measure, and why a shifted marker is not a result — is in How to Decide. What follows is what it means for this therapy in particular.

This matters because cell numbers are the most commonly advertised figure in this field.

Two products with identical cell counts can differ enormously in biological capability. A viable, metabolically fit cell carrying activating receptors is not equivalent to an exhausted cell, or one damaged by freezing and thawing.

In the CAR-NK trial described above, products with similar cell counts and surface markers performed differently depending on the quality of the source material — differences in metabolic fitness, persistence, and tumor control.

A figure like “70 million cells” is not enough to evaluate anything. In a 70 kg adult that is roughly one million cells per kilogram. Whether that is a lot or a little depends entirely on the product, its viability, its potency, and the protocol.

“Activated” is similarly uninformative on its own. The word does not tell you which cytokines were used, which receptors increased, how long the effect lasts, or whether the cells retain killing ability after infusion.

To actually assess a product you would need: the cell source; viability; purity (how many cells are genuinely NK cells rather than something else); residual T cell content; expression of receptors like CD16 and NKG2D; a potency assay (a laboratory test measuring whether the cells actually kill, not just that they exist); expansion duration; cytokines used; sterility and endotoxin testing; post-thaw recovery; and batch-to-batch consistency.

Immune markers are not the same as clinical benefit

An increase in NK cells in your blood does not demonstrate that they reached the tumor. Increased killing of a laboratory cell line does not demonstrate tumor reduction inside a patient.

The following are useful as exploratory markers, but none of them by itself proves benefit: higher NK cell counts, increased NKG2D expression, more interferon-gamma production, greater killing of cancer cells in a laboratory dish, higher perforin or granzyme levels, reduced circulating tumor cells, changed cytokine profiles, or shifts in isolated tumor markers.

All of these need to be connected to something that matters to you: imaging response, progression-free survival, overall survival, symptoms, or quality of life.

A therapy can measurably change your immune system without controlling your cancer.

About exosomes

Some providers offer exosomes alongside NK cells, described as activators.

“Exosomes” does not identify a specific product. They are extracellular vesicles whose contents and effects depend entirely on the cell they came from, how they were cultured, isolated and purified, their dose, and how they are administered.

They can stimulate or suppress immunity. There is no universal anti-tumor property attached to the term. Exosomes derived from tumors, for instance, can do the opposite — promoting immune escape and reducing NKG2D function.

NK-derived exosomes contain perforin and granzymes and have shown anti-tumor effects in cell culture and animal models. That evidence is predominantly preclinical. There is no robust clinical evidence that adding a product called “exosomes” to an NK infusion predictably activates those cells in patients, increases response rates, or improves survival.

If exosomes are offered to you, the questions are: from what cell source, isolated how, at what dose, with what human evidence.

Sources — two reviews. Kugeratski FG, Kalluri R. Exosomes as mediators of immune regulation and immunotherapy in cancer. The FEBS Journal, 2021. PubMed 32910536 · doi:10.1111/febs.15558

Hatami Z, et al. Natural killer cell-derived exosomes for cancer immunotherapy: innovative therapeutics art. Cancer Cell International, 2023. PubMed 37543612 · doi:10.1186/s12935-023-02996-6

Safety

NK therapies should not be described as free of side effects.

Reported events across trials include fever, chills, fatigue, infusion reactions, low blood counts, infections, and liver enzyme changes.

Much of the toxicity in these protocols comes from what accompanies the NK cells, not the cells themselves. Lymphodepleting chemotherapy (fludarabine and cyclophosphamide) causes neutropenia, anemia, low platelets and infection risk. IL-2 can cause vascular and inflammatory toxicity. The absence of severe direct NK toxicity does not make the overall protocol risk-free.

Genetically modified products can damage healthy tissue if the marker they were built to target also appears on normal cells — known as on-target, off-tumor toxicity.

What does appear favorable: in the small CAR-NK studies to date, cytokine release syndrome and neurotoxicity have been less frequent or less severe than historically seen with some CAR-T therapies. But there are no randomized comparisons large enough to claim general safety superiority.

And favorable safety is not evidence of benefit. A treatment that does nothing also has few side effects.

How the therapy is prepared

Autologous cells — meaning your own — come from your blood, drawn under specific conditions, then isolated and multiplied in a laboratory over roughly 10 to 12 days.

Allogeneic cells — meaning from someone else — come from a healthy donor, typically a family member. Where no suitable donor is available, cells can come from a bank of screened donors.

Autologous cells are not always viable. Whether your cells can be collected and expanded successfully depends on your condition and treatment history. This is a clinical determination, not a preference you select.

Each dose undergoes quality control before release, verifying cell count, purity, and NK proportion. Prepared doses have a viability window measured in hours, so scheduling depends on laboratory timing.

Dose and protocol are clinical decisions made by the treating physician after reviewing your case.

How to read the phase of a study

This is worth knowing for any treatment you evaluate, here or anywhere else:

• Phase I asks whether a treatment is safe and at what dose. Usually small, often without a comparison group. It cannot establish that a treatment works.
• Phase II looks for signals of activity in a specific condition. Still limited, sometimes uncontrolled.
• Phase III compares against standard treatment in a large population. This is what regulatory approval generally requires, and what shows whether patients actually live longer.

Most published NK cell research in solid tumors is phase I or early phase II. That describes where the research currently stands — it is not a verdict on the therapy.

When any website cites a response rate, ask which phase produced it and whether there was a control group. Without both, a percentage tells you very little — and that applies to this page as much as to any other.

Plausibility, activity, and benefit are three different things

This is the general test described in How to Decide, applied to this therapy specifically.

Biological plausibility means there is a credible mechanism. For NK cells this is strong — they demonstrably kill cancer cells in laboratory and animal models.

Clinical activity means responses have been observed in real patients. This also exists — some CAR-NK and memory-like NK products have produced remissions.

Established clinical benefit means sufficient, reproducible evidence that a therapy improves outcomes that matter compared with available alternatives. This level does not yet exist for NK cell therapy as a category.

Confusing these three is the single most common error in how these treatments are marketed. That a cell kills cancer in a dish does not mean an infusion extends a life. That some patients responded in a small study does not mean the treatment beats standard care.

Regulatory status

There is no FDA-approved NK cell therapy for cancer. As of 2026, the FDA’s official list of approved cellular and gene therapy products includes multiple CAR-T therapies, cord blood products and tumor-infiltrating lymphocyte therapy — but no NK cell product approved as a cancer treatment.

Orphan drug designation, fast track, breakthrough therapy status, an approved investigational application, or registration on ClinicalTrials.gov are not approvals. They indicate that development is underway.

Mexico regulates cell therapies under its own regulatory framework, which is different from the one used in the United States. A different framework is not an absence of one—but the two are not equivalent, and an authorization granted under one does not carry the meaning of an authorization granted under the other.

A therapy being authorized in one country and not another does not establish that it works. Different regulators apply different standards.

Source — regulator’s own list. U.S. Food and Drug Administration, Approved Cellular and Gene Therapy Products.

Claims that the evidence does not support

If you see any of these anywhere, the evidence does not back them:

• NK cells effectively treat any type of cancer
• NK cells eliminate cancer without affecting the patient
• NK cells are a proven alternative to chemotherapy
• Exosomes activate NK cells to improve treatment effectiveness
• A dose of X million cells guarantees an adequate response
• CAR-NK results demonstrate that NK therapy works
• NK cells produce results without toxicity
• NK cells prevent metastasis or recurrence
• NK cell therapy is approved for treating cancer

What to ask any provider offering this

These are the product-level questions, specific to this therapy. The general ones — about the process, the cost, and the care you are already receiving — are in How to Decide. Bring both sets to us and to anyone else.

About the product — What is the cell source? Autologous, allogeneic, cord-derived, cell line? Is it genetically modified? What is the viability and purity? Is there a functional potency assay, or only a cell count?

About the evidence — What published studies used this product? In which cancers? How many patients? What response rate, and measured how? Was there a control group?

About my case — What evidence supports using this for my specific diagnosis? What is the realistic expectation? What happens if it does not work?

About the protocol — How many infusions? Is lymphodepletion involved (chemotherapy beforehand to make room for the cells)? What other treatments are given at the same time, and how would we know what caused any change?

A provider who cannot answer these, or who answers vaguely, has told you something important.

What this means for you

No one can tell you from this page whether this is appropriate for your situation. That depends on your diagnosis, treatment history, current condition, and clinical factors only a physician reviewing your records can assess.

What this page can tell you is the shape of the evidence honestly:

• The biology is well established and the rationale is strong
• Real clinical activity has been demonstrated in selected blood cancers, with specific engineered products
• Safety is generally favorable, though it depends on the full protocol
• Effectiveness in solid tumors is not established
• There is no FDA-approved NK therapy for cancer
• Evidence of survival benefit is insufficient for most indications

Research use within regulated trials is scientifically justified. Presentation as a proven treatment, or as a substitute for standard care, is not supported by the available evidence.

If you are considering this, bring this page and these studies to your own oncologist. And do not change anything about your current treatment without talking to them first.

References

Every claim on this page that rests on a study is linked to it above. They are collected here so the list can be checked in one pass — by you, by your oncologist, or by anyone who wants to see whether this page says what its sources say.

Trials

1. Marin D, et al. Safety, efficacy and determinants of response of allogeneic CD19-specific CAR-NK cells in CD19+ B cell tumors: a phase 1/2 trial. Nature Medicine, 2024. PubMed · doi
2. Romee R, et al. Cytokine-induced memory-like natural killer cells exhibit enhanced responses against myeloid leukemia. Science Translational Medicine, 2016 — phase 1. PubMed · doi
3. Bednarski JJ, et al. Donor memory-like NK cells persist and induce remissions in pediatric patients with relapsed AML after transplant. Blood, 2022 — phase 1. PubMed · doi
4. Nguyen R, et al. A phase II clinical trial of adoptive transfer of haploidentical natural killer cells for consolidation therapy of pediatric acute myeloid leukemia. Journal for ImmunoTherapy of Cancer, 2019 — the trial where it did not work. PubMed · doi

Systematic review

5. Park H, et al. Efficacy and safety of natural killer cell therapy in patients with solid tumors: a systematic review and meta-analysis. Frontiers in Immunology, 2024. PubMed · PMC · doi

Reviews

6. Balkhi S, et al. CAR-NK cell therapy: promise and challenges in solid tumors. Frontiers in Immunology, 2025. PubMed · doi
7. Oh S, et al. Natural killer cell therapy: a new treatment paradigm for solid tumors. Cancers, 2019. PubMed · doi
8. Lizana-Vasquez GD, et al. The application of autologous cancer immunotherapies in the age of memory-NK cells. Frontiers in Immunology, 2023. PubMed · doi
9. Kugeratski FG, Kalluri R. Exosomes as mediators of immune regulation and immunotherapy in cancer. The FEBS Journal, 2021. PubMed · doi
10. Hatami Z, et al. Natural killer cell-derived exosomes for cancer immunotherapy: innovative therapeutics art. Cancer Cell International, 2023. PubMed · doi

Regulatory

11. U.S. Food and Drug Administration. Approved Cellular and Gene Therapy Products.

Note what is missing from this list, because it is the point of the page: there is no phase III trial showing that NK cell therapy improves survival in solid tumors. If one existed, it would be here.

This page provides general information and is not medical advice. It does not establish a physician-patient relationship, and it is not a representation that any treatment is appropriate for any individual. Decisions about your treatment should be made with your treating physician.