Safety of genetically modified T cells

Many classes of antineoplastic therapeutics, including radiation and chemotherapy, have the potential to generate off-target genomic mutations that lead to subsequent (or secondary) malignancies. Based on their antitumor efficacy and potential to induce long-term remissions, genetically modified immune effector cells (IECs) are one of the most exciting new classes of therapeutics for hematologic malignancies, and are in early development for treating solid tumors. In this issue of Blood, Steffin et al¹ report on the rates of subsequent malignancies in >1000 patient-years of follow-up across multiple clinical trials of genetically modified IECs for cancer, and reassuringly, found no increased risk of secondary cancers (see figure).

Although genetically modified IECs do not instigate nonspecific mutations in bystander tissues, the insertion of novel genes, or transgenes, into living cells carries the potential risk of insertional mutagenesis. In other words, it is possible for the transgene to disrupt a tumor-suppressor gene or for the promoter elements that drive the expression of the transgene to also drive the expression of a nearby oncogene in the modified cell. Some of the first clinical trials of gene addition were performed with retroviral vector–mediated gene transfer into human hematopoietic stem cells in children with immunodeficiency; unfortunately, several patients developed leukemia that was related to the integration of the vector and transgene.² Because of this potential risk for transgene or vector-driven malignancy, regulatory authorities recommend a minimum of 15 years of long-term follow-up in patients receiving any genetically modified cell therapy.³ 

Until now, there has been only 1 long-term study of patients treated with T cells modified with retroviral vectors, and that study showed that, in ∼540 patient-years of follow-up over the course of a decade, T cells transduced with first-generation CD4ζ chimeric antigen receptors (CARs) via retroviral vectors maintained expression and function of the transgene without clonal expansion of the transferred T cells.⁴ The landmark study by Steffin et al retrospectively reviews 340 adult and pediatric patients with cancer who received, via viral vectors, ≥1 doses of IECs that were genetically modified with a variety of transgenes. Of note, the transgenes included second-generation CARs with different structures and specificities, as well as transgenes meant to increase the persistence of T cells, such as a dominant-negative transforming growth factor (TGF)-β receptor. Because all the patients had recurrent or refractory cancer and were treated in early-phase clinical trials, they had had exposure to chemotherapy before enrolling in the IEC protocol. Remarkably, for a single center, the study population included patients treated with 27 different protocols, and all had received either autologous, donor-derived, or virus-specific T-cell products. The median follow-up was 14.9 months, with 56.5%, 21.2%, and 8.5% of patients having been observed for more than 1, 5, or 10 years, respectively. Overall, there were 1027 years of cumulative follow-up, and Steffin et al found that 3.8% of the patients developed secondary cancers. The rate of cumulative secondary cancers at 5 years in patients receiving retrovirus-modified IECs (3.6%) was the same as that in a control group of patients who received Epstein-Barr virus–specific T cells that were not genetically modified (3.6%), and both were the same as published rates of secondary malignancies induced by chemotherapy alone (2% to 5%).⁵ Notably, 75% of the subsequent tumors were solid tumors, with the most common being basal cell carcinoma. Those tumors would not be expected to be related to the genetic modification process or T-cell manipulations. A deeper look at the subsequent tumors indicated that none expressed the transgene introduced into the T cells, and no patients had developed a replication-competent retrovirus in the course of follow-up. Overall, the data demonstrate a negligible rate of genotoxicity by retroviral vectors used to modify T cells. Of note, patients treated with other immune or hematopoietic cells besides mature T cells were not examined, nor did any patients in this cohort receive treatments with cells that had been modified with gene-editing technologies, such as TALENS or CRISPR/Cas9.

To date, there have been no reported cases of secondary malignancies associated with T cells modified with lentiviral vectors, although clonal expansions have been reported in 2 case reports, in which CARs integrated into the CBL gene⁶ or the TET2 gene.⁷ Another recent study published in Blood reported on patients with B-cell lymphoma who were treated with CD19-directed T cells that were modified with a piggyBac transposon-based gene-delivery approach. Two of 10 patients treated developed T-cell lymphoma associated with CAR expression⁸; a deeper molecular investigation revealed that, although there was no transgene insertion into typical oncogenes, 1 patient’s CAR T-cell lymphoma had high transgene copy numbers (24 integration sites), and both CAR T-cell lymphomas had structural and copy number variants and dysregulation of transcripts surrounding the transgene.⁹ In contrast, Steffin et al noted that their mean vector copy number per transduced cell was 5.38 (range, 1.48-9.40), and most of the patients who had subsequent malignancies had detectable transgene for less than 12 months. It remains possible that when transgenes are present in higher copy numbers or if they successfully confer longer persistence of the modified T cells, they may still exert genotoxicity, regardless of the gene delivery system, such as those that are retrovirus, lentivirus, transposon, or nuclease based. Nevertheless, the work by Steffin et al represents the first large study of long-term follow-up of patients treated with retroviral vector-modified T cells, and the results are, notably, highly reassuring. In this light, it is reasonable to anticipate that retroviral vector modification of mature T cells causes genotoxicity rarely enough that relaxing the uniquely intensive and prolonged monitoring is warranted.