Leukemia is the most common type of cancer that occurs in childhood, and as recently as the early 1940s it was a death sentence. In his fervent search for treatments, Dr. Sidney Farber tested a litany of chemicals, some of which made his patients weaker and worsened their symptoms. But in 1947-48, there was a breakthrough when a few of his leukemic children went into remission after treatment with antifolates, the very first cancer drugs.

In the 69 years since the first clinical demonstration of chemotherapy, outcomes for most cancers have improved dramatically. Nowadays, blood cancers are generally among the more 'survivable' cancers. The type Farber's patients had, acute lymphoblastic leukemia (ALL), has an 85 percent survival rate. The minority of patients that don't respond to treatment are considered 'refractory,' and those whose cancer returns are 'relapsed.' These unlucky patients' only remaining option used to be a long-shot stem cell transplant.

But a newly approved treatment called CAR-T is already changing the calculus. It shows such promise that, just two months after the first CAR-T therapy for ALL was approved by the FDA, a second one for another type of blood cancer was approved Wednesday. But underlying the optimism, there's a reason CAR-T is reserved for patients that fail to improve under front-line treatments: it comes with horrid side effects and can itself be fatal. We are slowly finding exciting treatments for formerly deadly diseases, but the pathway remains riddled with landmines.

Blood cancers are liquid tumors, in which cancerous white blood cells outgrow healthy cells in the bone marrow and circulate throughout the body. ALL manifests in immature white blood cells called lymphoblasts. Normally, healthy lymphoblasts develop into two types of immune cells. The first are B cells, which produce proteins called antibodies that monitor the bloodstream for infectious agents. The second are T cells – the ‘T’ in CAR-T. While T cells play many roles, think of them as the generals of the immune system: when they get activated, they unleash havoc on their target.

All cells display molecules on their membranes called antigens that identify them to the antibodies as native or foreign. For example, if you’ve ever donated blood, you’re aware of the A/B/AB/O system for red blood cells. A type-A person’s cells display A antigen, and they have antibodies against B antigen; if they receive a mismatched transfusion of type-B blood, anti-B antibodies bind the B antigen and elicit a potentially fatal immune response against the foreign blood cells. This not only shows the importance of blood type characterization and matching for transfusions in emergencies, but also the potency of the immune response.

CAR-T harnesses that immune response, directing it against cancer cells.

Here’s how it works: oncologists isolate T cells from the patient’s own blood. Those T cells are grafted with DNA that encodes a 'chimeric antigen receptor' – the ‘CAR’ in CAR-T. It's an antibody fragment genetically engineered to recognize an antigen unique to the cancer cells - in this case, a B cell protein called CD19. Suitably primed, specialists grow more of the CAR-T cells in the lab before transfusing them back into the patient, where they marshal the immune system into action against the cancer.

CAR-T is not a drug; it’s a genetically modified cell therapy. That’s why it’s so innovative, but more important than innovation is effectiveness. In June, the pharmaceutical company Novartis released clinical data showing complete remission in 52 of 63 pediatric & young adult patients with relapsed or refractory ALL. The successfully treated patients remained in remission over six months after treatment, with no evidence of residual disease in their bone marrow. Convinced of its efficacy, the FDA approved this use of CAR-T treatment under the name Kymriah.

This therapy could be used against other B cell cancers that express CD19, including multiple myeloma and the majority of leukemias and lymphomas. In fact, Wednesday's approval of Gilead/Kite's anti-CD19 CAR-T is for relapsed & refractory non-Hodgkin's lymphoma. Cancer cells can become resistant to anti-CD19 CAR-T treatment by mutating or deleting the targeted antigen, a process called antigen-loss escape that is responsible for some relapses. But beyond CD19, there are already CARs against many other blood cell antigens in clinical trials.

Unlike liquid tumors, solid tumors embedded in organ tissue may be poorly accessible to circulating CAR-T cells. Upon arriving, the poorly oxygenated tumor microenvironment is like caltrops defending against the encroaching immune response. Furthermore, sold tumors are made up of many distinct cell types with different antigens, warranting multiple CARs to keep up the assault. Modifications such as oxygen-sensitive modules could optimize CAR-T treatment for solid tumors, but this is a remaining challenge for the field.

A bigger challenge for all CAR-T treatments is to improve safety. One of the ways T cells unleash havoc is by spraying out pro-inflammatory molecules, called cytokines, that recruit other immune cells to join the fray. If the CAR-T cells are too potent, or if there are too many of them, the patient is swamped in cytokines, resulting in a severe inflammatory condition called cytokine release syndrome (CRS). In the successful Novartis trial, nearly half of the patients experienced this side effect, though they all recovered. Most CRS sufferers experience a high fever that recedes in the two weeks following treatment.

But in the worst cases of CRS, which get dubbed cytokine storm, that fever can lead to a fatal swelling of the brain. Earlier this year, Juno Therapeutics halted one of its CAR-T trials in response to five cumulative deaths since last year. Before being acquired by Gilead, Kite Pharma also reported three trial deaths. The most recent death came in Cellectis’s CAR-T trial, nine days post-treatment. The CRS conundrum is a balancing act: a weak immune response will be insufficient to defeat the cancer, but overactivation causes toxic inflammation and death. If that balance can be struck to reduce the incidence of CRS, it could push CAR-T past chemotherapy and radiation to the forefront of treatment options.

The excitement around CAR-T may not yet be at its crescendo. As of this writing, nearly 400 CAR-T clinical trials are registered with the FDA, some even targeting brain tumors. But CAR-T is not some miracle cure. CRS is a potentially deadly side effect that many patients will suffer. And for those without means, successful remission may mean a lifetime of debt: the cost of the two approved CAR-T treatments, eachlisted in the hundreds of thousands of dollars, is expected to spiral over $1 milllion per patient.

But we also shouldn't be so jaded as to miss the astounding impact CAR-T is already having. Fittingly, childhood ALL was the proving ground for a risky new treatment in patients who'd run out of options. If only Sidney Farber could see us now.