Therapeutic plasma exchange is not an approved treatment for cancer, and no one should be offering it as one today.

The reason the question keeps coming up is that cancer defends itself partly by broadcasting molecules into the bloodstream, and those molecules are removable. Whether removing them helps patients is now being tested in controlled trials. The early results are interesting enough to be worth understanding.

Where plasma exchange is already established in cancer care

There is a narrow set of situations where exchange has a defined role, and it has nothing to do with attacking the tumor.

The clearest is symptomatic hyperviscosity from a monoclonal gammopathy. In Waldenström macroglobulinemia and some cases of multiple myeloma, the cancer produces so much abnormal immunoglobulin that the blood physically thickens. Patients develop visual changes, headache, confusion, and mucosal bleeding. Plasma exchange removes the paraprotein and the symptoms resolve within hours. It is one of the most immediate responses in all of apheresis medicine, and it appears in the American Society for Apheresis guidelines as a Category I indication, meaning apheresis is accepted as first-line therapy.

Several paraneoplastic neurologic syndromes, where a cancer provokes antibodies that attack the nervous system, also carry apheresis indications. In those cases the target is the antibody, not the tumor.

Both of these are established medicine. Neither is what people mean when they ask whether plasma exchange treats cancer.

The part that is genuinely new

For most of the history of oncology, the weapons pointed at the tumor. Surgery removed it, radiation burned it, chemotherapy poisoned it. Then checkpoint inhibitors arrived and changed the target: rather than attacking the tumor, they release the brakes the tumor had applied to the immune system.

They work remarkably well in some patients. Most patients, though, do not respond, and many who do eventually relapse.

Part of the reason appears to be that cancer does not only fight at the tumor's edge. Tumors shed PD-L1, the protein that switches off T cells on contact, into the circulation in two forms: a free-floating soluble version, and a version carried on the surface of exosomes, the small vesicles cells release as a kind of postal system. Both travel. The result is a systemic signal that dampens T cells far from the tumor, in lymph nodes and tissues the cancer has never reached.

There is a second consequence. A bloodstream full of circulating PD-L1 can act as a decoy, absorbing the antibody drugs designed to block PD-L1 before they arrive at the tumor. Higher circulating levels track with worse survival and poorer response to immunotherapy across multiple cancers.

That is the setup for the obvious question. If the signal is soluble and circulating, and plasma exchange exists to remove soluble circulating things, what happens if you take it out?

This has been tried before, and it went badly in an instructive way

In 1985, an oncologist named M. Rigdon Lentz built a machine to remove an immunosuppressive factor in cancer patients' blood without knowing what the factor was. Clinicians had known for years that serum from patients with advanced cancer blunted immune cells in a dish, and that whatever caused it was small enough to pass through a filter that held antibodies back.

The response in patients was hard to miss. Within hours of treatment, tumors became warm, swollen, and tender. Sequential biopsies showed lymphocytes gathering around the vessels feeding the tumor and then pushing into it, followed by patches of tumor cell death that merged into fields of necrosis. Among sixteen patients with metastatic solid tumors receiving no other cancer treatment, tumor inflammation appeared in fourteen and measurable disease fell by half or more in six.

Only later was the removed factor identified, and the answer explained the tumor inflammation.

Tumor necrosis factor is named for what it does. It is one of the signals the immune system uses to trigger cancer cell death, and it drives the inflammation that brings T cells into a tumor. Cells normally receive that signal through a receptor anchored in their surface membrane.

What the machine had been discarding was the outer arm of that receptor, snapped off the cell surface and circulating free. Floating loose, it still binds TNF perfectly well. It simply has no cell attached to it and nowhere to deliver the message. It is a decoy, and every molecule of TNF it captures is one that never reaches a cancer cell.

If that sounds like a laboratory curiosity, consider what the pharmaceutical industry does with the same molecule. A manufactured soluble TNF receptor, fused to an antibody fragment, is etanercept, given to patients with rheumatoid arthritis for the express purpose of mopping up TNF and shutting inflammation down. It works. Tumors had arrived at the same answer first and were producing their own supply. Patients with cancer carry two to three times the normal concentration of these shed receptors, and the excess tracks with the stage of their disease.

The most striking result in the series was also its most difficult. In three patients the tumors were destroyed so rapidly and so completely that the patients developed tumor lysis syndrome, a metabolic emergency caused by the contents of dying cancer cells flooding the bloodstream faster than the kidneys can clear them. At autopsy each showed near-total destruction of the cancer. All three died.

Tumor lysis syndrome is far better understood today than it was in 1985. It is anticipated, monitored for, and routinely managed in oncology practice, and it is an expected consideration whenever a treatment is capable of killing a large tumor burden quickly. What has not happened in the forty years since is a randomized trial of that approach, so the question of whether it extends survival remains open.

It is worth being precise about what that machine was. It was a filtration device with a 150 kilodalton cutoff, built to remove one class of small proteins. It was not plasma exchange, and plasma exchange has not been associated with tumor lysis syndrome.

The current trials

A group at the Mayo Clinic took up the modern version of the question. They first showed that plasma exchange physically clears these molecules, removing roughly seventy percent of soluble PD-L1 and a similar share of PD-L1-bearing vesicles per session.

Then came ReCIPE-M1 (NCT04581382), a phase I trial in eighteen patients with metastatic melanoma that was progressing despite checkpoint inhibition, which is about as difficult a population as exists in melanoma. Patients received radiation to a few tumor sites, three days of plasma exchange, and a re-challenge with checkpoint drugs. The treatment was safe. Soluble PD-L1 dropped substantially. Eleven of the eighteen had tumors shrink, three of them completely, for a response rate above sixty percent. How thoroughly a patient's PD-L1 was suppressed tracked with how long they survived.

This was a phase I trial designed to establish safety, with eighteen patients and no randomized control arm. Patients received three interventions at once, so the trial cannot isolate what the exchange itself contributed. And the finding that PD-L1 suppression tracked with survival is a correlation within the treated group. It is not a demonstration that exchange extended anyone's life compared with not receiving it.

What can be said is that a response rate above sixty percent, with complete responses, in melanoma that had kept growing through the best available immunotherapy, is well beyond what radiation and a re-challenge alone would be expected to produce in disease that refractory. That is not proof of causation. It is the pattern you would expect if the hypothesis were right, and it is enough to justify a controlled trial.

That trial is now running. ReCIPE-B1 (NCT07087860) is a randomized phase II at Mayo, testing plasma exchange alongside enfortumab vedotin and pembrolizumab in bladder cancer. It began accrual in August 2025 and is currently recruiting, with primary completion estimated for 2028. Randomized is the word that matters. It is the design that can answer the question the phase I could only raise.

Where this stands

There is a serious scientific idea here, being tested properly, by a serious institution, and the answer is not in yet.

Outside a trial, plasma exchange is not a cancer therapy, and this article is not a suggestion that anyone pursue it as one. If the mechanism looks relevant to your situation, the productive conversation is with your oncologist about whether a trial is open to you. ReCIPE-B1 is listed on ClinicalTrials.gov, and more trials in this area are likely as the field develops.

I write about therapeutic plasma exchange because I perform it, and part of that is being straight about where the evidence sits. For the established indications, it is settled medicine. For cancer immunotherapy, we have a mechanism that makes sense, a phase I signal that is hard to dismiss, and a randomized trial that should tell us whether it holds.

If you are trying to make sense of what you have read about plasma exchange, in oncology or anywhere else, you are welcome to get in touch. I am glad to help people understand what the evidence does and does not show, including when the answer is that we do not know yet.