Spinal muscular atrophy: Tumors reported following Novartis gene therapy

Novartis’ gene therapy Zolgensma is once again under scrutiny over potentially serious side effects. Since its initial approval in 2019, the Swiss pharmaceutical group has reported nine cases of tumors in previously treated patients. According to the company, two of these cases suggest that DNA from the viral vector used in the therapy had integrated into the patients’ genomes. In both cases, the patients had received treatment and remain under observation, Novartis said. According to the company, more than 6,000 patients have received Zolgensma through clinical trials, compassionate use programs and routine medical care.

Zoom in: The reports emerged in early November 2026, after Novartis had apparently first informed treating physicians. Patient organizations such as SMA Europe and Cure SMA subsequently called for transparent communication with affected families. On November 4, the British patient organization SMA UK and the country’s National Multidisciplinary Team reported that pathological examinations were already available for three of the nine tumor cases. Two tumors had reportedly been linked to Zolgensma treatment, while one had not. The remaining cases had not yet been conclusively assessed. The relevant regulatory authorities are investigating the available information.

A reliable risk assessment is not yet possible. Neither the tumor types nor the precise molecular findings and individual disease courses have been comprehensively documented publicly. Nor can a reliable incidence rate be calculated from the nine reports among more than 6,000 treated patients. Spontaneous postmarketing reports do not constitute a complete registry of all treated patients and do not allow direct conclusions about whether a drug caused the reported events.

How it works: Zolgensma (onasemnogene abeparvovec) is an in vivo gene therapy. It introduces a functional copy of the human SMN1 gene into cells. Spinal muscular atrophy is caused by a defect in this gene, which is needed to produce survival motor neuron (SMN) protein. This protein is essential for the survival and function of motor neurons.

The delivery vehicle is a genetically engineered, nonreplicating adeno-associated virus serotype 9 (AAV9). It transports the additional gene copy into target cells, where it is intended to compensate for the missing function.

A distinctive feature of AAV vectors is that the introduced genetic material remains predominantly outside the chromosomes in the cell nucleus, in structures known as episomal DNA. In other words, it is not normally deliberately integrated into the human genome. This is a fundamental difference from certain other gene therapy approaches.

Yes, but: this separation is not absolute. AAV vector DNA can also integrate randomly into the human genome. Such integration events have been described in the scientific literature and are explicitly mentioned in the U.S. prescribing information for Zolgensma. The FDA acknowledges that the clinical significance of individual integration events remains unknown. However, they could theoretically increase the risk of tumor formation.

The biological mechanism behind this is known as insertional mutagenesis. If genetic material is inserted into the genome at an unfavorable location, it can, for example, alter the activity of neighboring genes. These could include genes that regulate cell growth or prevent uncontrolled cell division. If such control mechanisms are disrupted, they can contribute to tumor development over time.

This does not mean that every integration event is automatically dangerous. Among the determining factors are the location in the genome, the affected cell, the type of integration and any subsequent changes. However, the newly reported cases make clear that the theoretical risk cannot simply be dismissed as a purely academic concern.

Backstory: The latest debate is not the first serious safety controversy surrounding Zolgensma. In August 2022, Novartis reported two deaths following acute liver failure. The children had been treated in Russia and Kazakhstan and died approximately six to seven weeks after their respective infusions. According to the company’s subsequent safety communication, liver toxicity developed following the initiation of corticosteroid tapering.

Acute liver injury was already a known potential side effect. However, the two deaths prompted Novartis, in coordination with regulatory authorities, to update the product information. For the first time, it explicitly warned of fatal acute liver failure.

Reality check: The underlying mechanism, however, is likely different from that of the tumors now under discussion. Liver toxicity is primarily associated with the body’s immune response to the viral vector. The immune system recognizes components of the vector or the cells it has entered and can trigger a pronounced inflammatory response. This reaction can become particularly problematic when high doses of the vector are administered systemically.

The U.S. prescribing information therefore includes a warning about severe liver injury and acute liver failure, including fatal cases. It requires liver function testing before treatment, concomitant corticosteroid therapy and close monitoring for at least three months after infusion. These precautions are also detailed in a 2023 safety communication from the European Medicines Agency. Thrombotic microangiopathy and changes in platelet counts are also among the documented risks.

The two safety concerns therefore have different causes: While liver toxicity is primarily attributed to immune reactions, the tumor reports involve the possible integration of vector DNA into the genome. What they have in common is that both arise from the particular characteristics of a gene therapy that introduces large quantities of genetic material into the body and is intended to produce a long-lasting effect.

The big picture: The cases inevitably raise the question of whether they reveal a fundamental problem with gene and cell therapies. The answer is more nuanced: There are shared challenges, but the risks differ considerably depending on the technology, vector, target cell and treatment approach.

With AAV-based gene therapies such as Zolgensma, the primary concern is the immune response to the vector. There is also the rare but biologically plausible risk of vector DNA integration. Other gene therapies, by contrast, deliberately use integrating vectors.

One example is CAR-T cell therapy. In this approach, a patient’s own immune cells are collected and genetically modified so they can recognize and attack cancer cells. Lentiviral or retroviral vectors are frequently used to introduce the genetic instructions, with their genetic material deliberately integrated permanently into the T cells’ genomes.

Here, too, integration at an unfavorable location can affect the regulation of cell growth and division. In rare cases, this can lead to malignant transformation of the treated immune cells. In April 2024, the U.S. Food and Drug Administration (FDA) responded by strengthening the warnings for all autologous CD19- and BCMA-directed CAR-T cell therapies approved at the time. Since then, prescribing information has been required to explicitly warn of the risk of secondary T-cell malignancies. The FDA recommends lifelong monitoring of treated patients.

However, this concern cannot be generalized to all cell therapies. Not every cell therapy involves genetic modification, and not every gene therapy uses integrating vectors. The risks associated with genome editing using CRISPR/Cas also differ from those of conventional gene addition. For genome editing, the FDA’s safety guidance focuses, among other things, on unintended changes at off-target locations in the genome and structural alterations to chromosomes.

What many of these approaches have in common, however, is that their effects are intended to last for a long time. Unlike a conventional drug, whose active ingredient is broken down after a certain period, a genetic modification that has already taken place cannot easily be reversed.

What’s next: This is precisely what makes the safety assessment of these new therapies so challenging. Some risks can emerge immediately after treatment, for example through immune reactions or acute organ damage. Other potential consequences may only become apparent years later.

The FDA therefore published specific guidance on long-term follow-up after gene therapy as early as 2020. It notes that genetic modifications and vectors that persist in the body over long periods can cause delayed and potentially difficult-to-predict adverse effects. Depending on the product and its risk profile, long-term follow-up programs may be necessary.

The situation is particularly sensitive for Zolgensma: The treatment is administered predominantly to very young children, whose life expectancy can be substantially improved by the therapy. At the same time, they may have to live with the consequences of a one-time genetic intervention for decades.

The latest tumor reports therefore do not automatically call the benefits of the therapy into question. However, they highlight that the benefit-risk assessment of a gene therapy does not end with regulatory approval. It must be continually reassessed as the number of treated patients grows and longer periods of follow-up become available.