Dr Steven Rosenberg is pursuing a difficult extension of cancer immunotherapy: making a person’s own immune cells effective against more of the common cancers that form solid tumours. The ambition reaches beyond celebrating an earlier breakthrough. It concerns patients whose cancers have continued to grow despite several treatments and whose next options remain limited. At the U.S. National Cancer Institute, Rosenberg’s programme brings laboratory selection of immune cells into clinical trials. Its most interesting feature is the effort to make personalisation experimentally useful: identify cells that recognise an individual tumour, increase their numbers, and test whether they can produce a meaningful response. A long pursuit with a current obstacle NCI’s current profile lists Rosenberg as chief of its Surgery Branch. His work helped establish that manipulating the immune system can cause cancer regression. An NIH historical account describes the early investigation of interleukin-2 and the development of approaches using tumour-infiltrating lymphocytes, usually called TILs. These are immune cells already found within a tumour. Their presence creates an opportunity, but does not establish that every cell collected will attack cancer effectively. NCI’s patient guidance explains that researchers can test and select lymphocytes that recognise tumour cells, then expand them in the laboratory before infusion. This is the research problem behind the personalised label. The task involves identifying useful activity and preserving it through a demanding treatment process. Simply giving more cells is not the same as giving cells with the right recognition capabilities. Selecting the cells that matter In April 2025, Rosenberg and colleagues reported a phase 2 study involving 91 people with metastatic gastrointestinal cancers. Participants had previously received a median of four treatment regimens. The programme examined successive approaches, including unselected TILs, selected TILs, and selected TILs combined with pembrolizumab. NCI reported no objective responses among 18 recipients of unselected cells. Responses occurred in three of 39 people receiving selected TILs and eight of 34 receiving selected TILs plus pembrolizumab. The combination result was 23.5%, a signal worth investigating in cancers that had proved difficult to treat. The groups were sequential parts of a developing programme, rather than a simple randomised comparison establishing the contribution of each change. Tumour shrinkage also differs from a demonstrated survival advantage. These limits belong beside the result, particularly in a profile of a researcher associated with major advances. The next breakthrough is more specific The team’s stated next direction includes improving identification of cells that recognise several tumour-specific neoantigens. These are targets associated with a cancer’s mutations. The aim is to increase the proportion of patients who respond, rather than assuming that the existing approach already works broadly. A separate NCI report from July 2024 described another route: transferring personalised, tumour-reactive T-cell receptors into a patient’s blood-derived immune cells. In a small colorectal-cancer study, three of seven patients responded. The result supplied proof of possibility and a reason for refinement, not a universal solid-tumour treatment. These approaches share a question about recognition. Can researchers identify the molecular features that make a particular cancer vulnerable, then create a cell product able to act on that information? The answer must come from reproducible clinical results across appropriately designed studies. Why the difficult parts matter Cell therapy is an intensive pathway involving collection, laboratory work, preparatory treatment and specialist monitoring. NCI’s guidance describes weeks of cell growth and treatment to reduce competing immune cells before infusion. This places manufacturing and patient fitness within the scientific challenge, rather than treating them as administrative details. Safety is equally central. The 2025 report describes serious side effects in 30% of patients treated with selected TILs. A stronger antitumour response would have to be considered alongside those harms and the experience of undergoing the full regimen. Existing approvals must also be kept separate from experimental extensions. Lifileucel has an approved role in advanced melanoma. That does not establish approval of the personalised gastrointestinal-cancer approaches described here. A familiar therapeutic family can contain both established treatments and early research programmes. A researcher to follow for the next question Graymedy’s reading of Rosenberg’s work is that the pursuit remains compelling because it keeps narrowing a large ambition into testable decisions: which target, which cells, which combination and which patients. Progress depends on improving those decisions together. The potential breakthrough would be a reliably effective, tolerable and deliverable personalised treatment for more solid cancers. Rosenberg and his collaborators have produced signals that justify continuing that search. Its unfinished nature is precisely what makes this a research story worth following. Sources: NCI: current investigator profile; NIH Catalyst: historical research profile; NCI: gastrointestinal-cancer TIL trial, April 2025; NCI: personalised TCR trial, July 2024; NCI: cell-transfer treatment context. Photo: Rhoda Baer / National Cancer Institute, 2008. Reviewed 4 November 2026. Independently written from public sources; no interview was conducted. General information, not individual medical advice.